THE ROLE OF WOMEN IN FORMER REBEL PARTIES By Elizabeth Lynn Brannon A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Political Science—Doctor of Philosophy 2021 ABSTRACT THE ROLE OF WOMEN IN FORMER REBEL PARTIES By Elizabeth Lynn Brannon In post-conflict environments, former rebel groups that transition into political parties are critical political actors, as their participation within formal politics is believed to be a crucial guarantee of continued peace and security after war. Yet, for many former rebel parties, the path to political success is littered with challenges in building substantial public support. This dissertation examines a salient strategy through which rebel parties build such support: the inclusion of women. I explore the role of women in rebel parties by asking three related questions: (1) How does women’s political representation in rebel parties compare to other political parties? (2) How do legacies of women’s wartime inclusion effect women’s representation in rebel parties? (3) Which women do rebel parties elect? I begin in Chapter 1 by comparing the role of women in rebel parties to other polit- ical parties in post-conflict environments. I argue that these parties have unique incentives to promote the representation of women. By including women, rebel parties stand to dis- tance themselves from their violent conflict legacies and demonstrate they are committed to upholding transparency, democracy, justice, and peace. Novel data on women’s candidacy and electoral success at the party-level in post-conflict Africa from 1970-2020 are presented. Findings suggest that rebel parties run and elect significantly more women than other politi- cal parties. I find in instances of ongoing violence, women’s political representation generally falters, suggesting that these political strategies are only effective during peace time. Chapter 2 explores the effect of conflict legacies on women’s inclusion in rebel parties. Rebel groups that transition into political parties are influenced by their conflict legacies in a myriad of ways. In particular, rebel parties adhere to practices that shaped their identity and garnered wartime support. I argue that the wartime inclusion of women in rebel groups is one such wartime behavior that will carry over to the post-war environment. Using a second novel dataset on women’s representation in a global sample of rebel parties from 1970-2020, I find that former rebel parties that included female combatants are more likely to elect a higher proportion of women after war, particularly when women’s presence serves to create similar benefits that women offer during war. The results show that when women’s wartime combat participation cultivated greater civilian aid and external support during war, rebel parties are more likely to elect women after war. Finally, Chapter 3 considers the role of former rebel women in their respective rebel parties. I ask how former rebel parties create post-conflict political opportunities for their female former members. I argue that the party’s integration of these women depends on women’s general contributions during war, particularly their roles in the group and its impact on group success. I present a third novel dataset on the election of former female rebels in a global sample of former rebel parties from 1970-2020 and find that women’s roles during conflict influence their post-conflict election. Moreover, their integration also influences greater recruitment of other non-rebel women. These findings demonstrate that while many former rebel women are marginalized after war, a select few are successful in translating their wartime experiences into political gains. Overall, this dissertation demonstrates the salient role of women in rebel parties, a role that has been consistently overlooked in the literature. Moreover, these findings suggest that the inclusion of women is a political strategy used by rebel parties, casting doubt on these parties’ actual commitment to women and peace. Copyright by ELIZABETH LYNN BRANNON 2021 To my mom and dad. v ACKNOWLEDGEMENTS As I reflect on my time at Michigan State, the saying “It takes a village . . .” feels remarkably appropriate, as I am not sure I would have made it to this point without the help of many, many generous mentors, colleagues, and friends. Foremost, I must thank my mentors at James Madison University, who inspired me to become an academic in the first place. To Jared Diener and Phil Frana, thank you for teaching me the rewards of taking intellectual risks and for showing me how to have fun while doing so. To Melinda Adams, thank you for the endless kindness, care, encouragement, and support that you offered. During my time at Michigan State, I have been lucky enough to have many supportive colleagues. Thank you to Caleb Lucas, Chrissy Scheller, Steve Anderson, and Max Welch for many lunches spent discussing work and life. To my office mate and friend, Kesicia Dickinson, I am grateful to have had you by my side as we navigated graduate school. To Jonathan King, Amy King, and Shane Wery, thank you for helping me get through difficult times (and always with a sense of humor) and for making grad school more enjoyable. I received a plethora of support during my graduate career that was crucial in the development of this dissertation. I am grateful to Cory Smidt and Connie Schibber for offering kind and thoughtful support early in my time at MSU. Thank you to the United States Institute of Peace and the Minerva Institute for the generous financial support that allowed me to finish this dissertation in such a timely manner. Thank you as well to Michigan State University’s Center for Gender in a Global Context, the Michael Bratton Field Work in Africa Fund, the College of Social Science, and the Department of Political Science for providing the funds that made this research possible. I am especially appreciative of Sarah Reckhow and all the support that she has offered during her time as Director of Graduate Studies. I also want to thank John Ishiyama, Mimmi Söderberg Kovacs, Gyda Sindre, Mette Kjaer, Merete Bech Seeberg, Ragnhild Muriaas, Sarah Brierley, and Shane MacGhiollabhui for their comments and feedback as I developed the proposal for my dissertation. vi I am eternally grateful to my dissertation committee, including Jakana Thomas, Jef- frey Conroy-Krutz, Michael Wahman, and Shahryar Minhas. Thank you so much for all the time, support, and encouragement you offered me. To Shahryar, thank you for your consis- tent support and detailed feedback, as well as for ensuring that my methodology sections are always sound. To Michael, without your suggestion to look into these “rebel parties,” I am sure this dissertation would have looked quite different. Thank you for encouraging me to develop my ideas, helping me to turn those ideas into research designs, and offering opportu- nities for me to share my work. To Jeff, I am so grateful for your extremely thoughtful and detail-oriented feedback and for your patient help as I tried to develop a plan for fieldwork. Finally, to my advisor, Jakana, I have so much to thank you for. During my time at MSU, you empowered me to become an independent scholar, while also offering endless support and encouragement. Thank you for reading too many drafts of everything I write and for thoughtfully engaging with all of it. Thank you for helping me to navigate the department and the discipline and for teaching me to be a better writer, researcher, and community member. You pushed me when I needed it, while ensuring that I knew I had your support and confidence. I am so grateful for your mentorship. I want to thank my friends and family for their constant love and encouragement, particularly throughout the last several years. To my best friends—Becca, Lauren, Brooke, Lea, Sarah, Rachael, and Hannah—thank you for never letting the distance feel too far and for making sure I remember that love and community are the true treasures in life. To my two big sisters, Mary Kate and Adrianna, thank you for showing me how to be brave and bold in pursuit of my ambitions. To my parents, I owe everything. Knowing you will catch me if I fall allows me to take chances and be courageous. Finally, thank you to my partner, Nico. Your support made writing this dissertation easier. You make me a better person (and academic!). Thank you for reading every draft and for listening to every new idea. You remind me not to take everything so seriously and make me laugh when I need it most. I am so thankful for your endless support and love. vii TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 0.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 0.2 The Questions and Argument . . . . . . . . . . . . . . . . . . . . . . . . . . 2 0.3 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 CHAPTER 1 WOMEN’S POLITICAL REPRESENTATION IN AFRICAN REBEL PARTIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.2 Women’s Representation in Post-Conflict States . . . . . . . . . . . . . . . . 10 1.3 Variation in Conflict Legacies in Post-Conflict Politics . . . . . . . . . . . . . 13 1.4 Strategic Benefits of Increasing Women’s Political Representation . . . . . . 14 1.5 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 1.5.1 Dependent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 1.5.2 Explanatory Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 22 1.5.3 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 1.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 1.6.1 The Effects of Conflict-Specific Contexts . . . . . . . . . . . . . . . . 27 1.6.2 Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 1.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 CHAPTER 2 LABORED LEGACIES: THE POST-CONFLICT IMPLICATIONS OF WOMEN’S WARTIME PARTICIPATION . . . . . . . . . . . . 35 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 2.2 Rebel Party Politics and Conflict Legacies . . . . . . . . . . . . . . . . . . . 38 2.3 Why Rebel Parties Elect Women . . . . . . . . . . . . . . . . . . . . . . . . 41 2.3.1 Women’s Inclusion and Ideology . . . . . . . . . . . . . . . . . . . . . 43 2.3.2 Women’s Inclusion and Domestic Support . . . . . . . . . . . . . . . 45 2.3.3 Women’s Inclusion and International Support . . . . . . . . . . . . . 47 2.4 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 2.4.1 Dependent Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.4.2 Independent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.4.3 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 2.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.5.1 Alternative Explanations . . . . . . . . . . . . . . . . . . . . . . . . . 57 2.5.2 Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 CHAPTER 3 THE ELECTION OF FORMER REBEL WOMEN . . . . . . . . . . 67 viii 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 3.2 The Election of Former Rebels . . . . . . . . . . . . . . . . . . . . . . . . . . 70 3.3 The Political Viability of Female Rebels . . . . . . . . . . . . . . . . . . . . 72 3.4 The Election of Former Female Rebels . . . . . . . . . . . . . . . . . . . . . 75 3.4.1 The Effects of Female Rebels’ Wartime Contributions . . . . . . . . . 76 3.5 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 3.5.1 Dependent Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 3.5.2 Explanatory Variables . . . . . . . . . . . . . . . . . . . . . . . . . . 88 3.5.3 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 3.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 3.6.1 Conflict and Post-Conflict Factors . . . . . . . . . . . . . . . . . . . . 94 3.6.2 Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 3.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.1 Summary of Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.2 Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 4.3 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 APPENDIX A APPENDIX TO CHAPTER 1 . . . . . . . . . . . . . . . . . . 106 APPENDIX B APPENDIX TO CHAPTER 2 . . . . . . . . . . . . . . . . . . 117 APPENDIX C APPENDIX TO CHAPTER 3 . . . . . . . . . . . . . . . . . . 129 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 ix LIST OF TABLES Table 1.1: Average Percent of Women Candidates and MPs by Party . . . . . . . 23 Table 1.2: Multilevel Linear Regression Results . . . . . . . . . . . . . . . . . . . 25 Table 2.1: Tobit Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Table 3.1: Tobit Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Table A1: Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Table A2: Multilevel Linear Regression Model (Time) . . . . . . . . . . . . . . . 107 Table A3: Multilevel Linear Regression Model (Candidacy & Conflict Outcome) . 108 Table A4: Multilevel Linear Regression Model (Electoral Success & Conflict Out- come) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Table A5: Multilevel Linear Regression Model (Ongoing Violence) . . . . . . . . 110 Table A6: Multilevel Linear Regression Model (Women’s Participation in Rebel Groups) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 Table A7: Multilevel Linear Regression Model (Alternate Democracy Operational- ization) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Table A8: Multilevel Linear Regression Model (Split Sample) . . . . . . . . . . . 113 Table A9: Multilevel Linear Regression Model (Alternate Gender Quotas Opera- tionalization) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Table A10: Multilevel Linear Regression Results (Seat Share) . . . . . . . . . . . . 115 Table A11: Tobit Model (Alternate Specification) . . . . . . . . . . . . . . . . . . 116 Table B1: Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Table B2: Tobit Model (Voter Perceptions) . . . . . . . . . . . . . . . . . . . . . 118 Table B3: Tobit Model (Seat Share) . . . . . . . . . . . . . . . . . . . . . . . . . 119 Table B4: Tobit Model (Alternate Operationalizations) . . . . . . . . . . . . . . 120 Table B5: Multiple Regression Analysis (Alternate Specification) . . . . . . . . . 121 Table B6: Zero Inflated Beta Regression (Alternate Specification) . . . . . . . . . 122 Table B7: Heckman Selection Model (Alternate Specification) . . . . . . . . . . . 123 x Table B8: Tobit Model (Conflict Characteristics) . . . . . . . . . . . . . . . . . . 124 Table B9: Tobit Model (Rebel Governance) . . . . . . . . . . . . . . . . . . . . . 125 Table B10: Tobit Model (Time) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 Table B11: Tobit Model (Conflict Outcome) . . . . . . . . . . . . . . . . . . . . . 127 Table C1: Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 Table C2: Tobit Model (Conflict Factors) . . . . . . . . . . . . . . . . . . . . . . 131 Table C3: Tobit Model (Post-Conflict Factors) . . . . . . . . . . . . . . . . . . . 132 Table C4: Tobit Model (Party Strength) . . . . . . . . . . . . . . . . . . . . . . . 133 Table C5: Tobit Analysis (Alternate DV) . . . . . . . . . . . . . . . . . . . . . . 134 Table C6: Tobit Model (Alternate Operationalizations I) . . . . . . . . . . . . . 135 Table C7: Tobit Model (Alternate Observations II) . . . . . . . . . . . . . . . . . 136 Table C8: Linear Regression Model (Alternate Specification) . . . . . . . . . . . 137 xi LIST OF FIGURES Figure 1.1: Estimated Proportion of Women Candidates . . . . . . . . . . . . . . 26 Figure 1.2: Estimated Proportion of Women Winners . . . . . . . . . . . . . . . . 26 Figure 1.3: Marginal Effect of Time . . . . . . . . . . . . . . . . . . . . . . . . . 28 Figure 1.4: Marginal Effect of Ongoing Violence . . . . . . . . . . . . . . . . . . . 28 Figure 2.1: Marginal Effect of Female Combatants . . . . . . . . . . . . . . . . . 55 Figure 2.2: First Differences of Female Combatants and Ideology . . . . . . . . . 56 Figure 2.3: First Differences of Female Combatants and Civilian Support . . . . . 57 Figure 2.4: First Differences of Female Combatants and External Support . . . . 58 Figure 2.5: Marginal Effect of Seat Share and Female Combatants . . . . . . . . . 60 Figure 2.6: Marginal Effect of Opposition Parties and Female Combatants . . . . 61 Figure 3.1: Proportion of Women’s Seats . . . . . . . . . . . . . . . . . . . . . . . 85 Figure 3.2: Prop. Former Female Rebels (Party) . . . . . . . . . . . . . . . . . . 86 Figure 3.3: Prop. Former Female Rebels (Women’s Seats) . . . . . . . . . . . . . 86 Figure 3.4: Marginal Effect of Female Combatants . . . . . . . . . . . . . . . . . 92 Figure 3.5: Marginal Effect of the Proportion of Female Combatants . . . . . . . 92 Figure 3.6: Marginal Effect of Women’s Access to State Jobs . . . . . . . . . . . . 93 Figure 3.7: Marginal Effect of Women’s Civil Society Participation . . . . . . . . 94 Figure 3.8: Marginal Effect of Time . . . . . . . . . . . . . . . . . . . . . . . . . 96 Figure A1: Marginal Effect of Female Combatants and Time . . . . . . . . . . . . 128 Figure A2: Marginal Effect of Female Combatants and Rebel Victory . . . . . . . 128 xii INTRODUCTION 0.1 Motivation Post-conflict environments are often defined by change and uncertainty. These two elements are intimately related, with some changes sparking greater uncertainty and others offering comfort and security. Former conflict actors, including rebel groups, are a consider- able source of this uncertainty, particularly when they are legally permitted to participate in formal politics after war, either through a negotiated agreement or by emerging victorious in the armed conflict. Citizens and government officials alike must stand by to see whether these former warmongers will adapt to institutionalized politics, or if they will continue pol- iticking through violence. Rebel groups transitioning into political parties (“rebel parties”), then, are tasked with taking hold of these moments of change to foster feelings of certainty. To be successful, they must convincingly signal that they are in fact committed to peace. Gains in women’s political representation are one such change that often strikes con- tentment and alleviates such uncertainty in post-conflict contexts, as women’s presence is empirically and colloquially associated with peace (Best, Shair-Rosenfield, and Wood 2019; Demeritt, Nichols, and Kelly 2014; Shair-Rosenfield and Wood 2017). Subsequently, women’s inclusion becomes a fix for the uncertainty that follows conflict (Berry 2015b, 2018a; Tripp 2015). It is a fix that many former rebel leaders have relied heavily on, using women as a scapegoat for peace despite these parties’ violent pasts (Berry and Lake 2017). In Rwanda, Berry (2015b) argues that President Kagame, former rebel leader of the RPF, has relied on the extraordinary levels of women’s political representation to serve as “a smoke screen” for their continued violence. In Uganda, women in government have similarly been used to expand support and to convince supporters that the former rebel party is less abusive and 1 corrupt than regimes of the past (Goetz 2002). However, beyond these and other similar case examples, there is not a broad sense of what the role of women is in former rebel party politics. Thus, this is the intention of this dissertation—to understand the role that women play in former rebel parties globally. 0.2 The Questions and Argument I address this gap with three primary questions: (1) How does women’s political repre- sentation in rebel parties compare to other political parties? (2) How do legacies of women’s wartime inclusion effect women’s representation in rebel parties? (3) Which women do rebel parties elect? Throughout each of the chapters that correspond to these questions, I highlight how the conflict legacies of rebel parties influence their post-conflict integration of women. In essence, the integration of women into these parties is a strategic political tool. By in- cluding women, I argue, rebel parties stand to distance themselves from their violent conflict legacies and demonstrate they are committed to upholding transparency, democracy, jus- tice, and peace. By integrating women at high levels, these parties can appropriate feminine stereotypes for their benefit. In the three chapters presented, I examine a different piece of this argument. Through each of these chapters, I demonstrate that women’s integration into rebel parties is strategic. Women are included when it serves an end for rebels—specifically when it serves to gain support and success. Chapter 1 considers how women’s political representation is distinct in former rebel parties, as compared to other political parties. I argue that while all parties could bene- fit from the heuristic shortcuts offered by including women and the stereotypes associated with them, these benefits will be especially pertinent for former rebel parties. The unique relationship between former rebel parties and violence, as well as their history of relying on non-conventional tactics to reach political goals, demands that rebel parties swiftly demon- strate that they have turned over a new leaf and are now committed to peace. To test this argument, I collected novel data on women’s representation at the party-level in Africa from 2 1970-2020. I find that rebel parties run and elect more women on average than all other political parties. However, this relationship is dependent on the state of peace, as when there is still ongoing conflict, the association between rebel parties and the election of women is no longer significant. Chapter 2 demonstrates that for many rebel parties, the integration of women is a continuation of wartime strategy. I ask how women’s participation in conflict influences their long-term political inclusion. Rebel party behavior is influenced by a number of wartime legacies, particularly those that shaped their identity and garnered wartime support. After war, rebels must recruit many new members to run for party seats, including women who were not originally members of the group during conflict. I argue that the wartime inclusion of women in rebel groups is one such wartime behavior that will carry over to the post-war environment, though not all of these women will be former rebel women. Using a second novel dataset on women’s representation in a global sample of rebel parties from 1970-2020, I find that former rebel parties that included female combatants are more likely to elect a higher proportion of women after war, particularly when women’s presence serves to create similar benefits that women offer during war. The results show that when women’s wartime combat participation cultivated greater domestic and international support during war, rebel parties are more likely to elect women after war. These findings offer nuance to understandings of how women’s wartime contributions effect their political standing after conflict. Finally, Chapter 3 examines which women rebel parties elect, specifically if and when former rebel women are elected. The literature on former female rebels often emphasizes the stark marginalization and discrimination these women experience after war for their deviation from traditional gender norms. However, when former rebel groups transition into political parties, they often recruit from their former membership. I argue that through their former rebel groups, former female rebels could theoretically overcome bias and marginalization to take on formally elected roles in government, as their male comrades do, and consider when they are most likely to be successful. I argue that former rebel groups elect a greater 3 proportion of former rebel women when rebel women served in elite roles within the group, contributed to wartime victory, and face lower levels of competition from other possible female recruits. I present a third novel dataset on the election of former female rebels in a global sample of former rebel parties from 1970-2020 and find that women’s roles during conflict influence their post-conflict election. I find additional evidence, however, that higher levels of qualified non-rebel women are associated with lower levels of election for former female rebels. This chapter demonstrates that a select few former rebel women do in fact build on their wartime experiences to enter politics after war. Finally, I end the dissertation with a conclusion chapter. In this, I discuss the impli- cations of my findings, as well as offer thoughts about how future research can build on this body of work. 0.3 Contributions These findings offer several important contributions to our understanding of former rebel party politics, women’s political representation after war, and the legacies of women’s engagement in conflict. Additionally, this dissertation includes three novel data collection efforts. First, the intersection of these research areas sheds light on the relevance of gender to rebel party politics, a previously under-explored area of study. Moreover, it demonstrates the centrality of gender to the function of rebel party politics. Previous work on rebel parties has only explored nascent changes to group identity. Rather, much of the work has focused on how groups adapt to formal politics while maintaining their wartime supporters. Further, this work also offers a more nuanced understanding of how former rebel parties gain political power after conflict. Many rebel parties have come to hold prominent positions in legislative politics. These studies demonstrate how women’s inclusion can be used as a tool to ascend to such positions. Finally, and perhaps most pertinently, the findings in these chapters show women’s association to peace can be and is manipulated. When these associations are manipulated in order to distract from violent legacies, and sometimes present violence, 4 there is reason for concern. Scholars have long demonstrated that while higher levels of women’s inclusion are associated with peace, women too can be as violent and corrupt as men. Women members of rebel parties are unlikely to actually pacify these groups, though they may have convinced domestic and international communities otherwise. Second, this work offers additional nuance to the study of women’s political repre- sentation after war. Previous work has attributed gains in women’s post-conflict political representation to changing gender norms, institutional changes, and pressure from domestic and international actors (Tripp 2015). A smaller body of work has demonstrated how in a number of cases, gains in women’s political representation is manipulated, fostered by po- litical leaders in order to reap domestic and international clout (Berry 2015b, 2018b). This dissertation adds to these bodies of work, demonstrating further the ways in which women’s political representation can be highly manipulated, tying this to wartime legacies. Addition- ally, I connect these developments to party politics. Previous studies of women’s political representation after war have generally been conducted at the aggregate level, explaining national variation. I theorize about how different parties have varying incentives after war to elect women. Moreover, the findings demonstrate that rebel parties may in many cases be actually spearheading these gains in the first place. Finally, the conclusions of this work offer further reason to be cautious about the actual gains for women that result from increases in women’s political representation. Scholars have argued that these post-conflict increases in representation for women are often not translated into real policy change and that when it is, it often only benefits a select few women (Berry 2015b, 2018a; Berry and Lake 2017; Goetz 2002; Tripp 2015). By demonstrating a further way in which women’s political represen- tation is manipulated, this dissertation suggests even greater caution in equating women’s presence with women’s influence. Third, this dissertation speaks to the legacies of women’s participation in civil con- flict. The literature on gender and conflict frequently reproduces two competing ideas. The first is the idea that women’s participation in conflict offers the opportunity for women to 5 “prove” themselves and thus can lead to changing ideas about gender roles, ultimately pro- ducing positive change for women. Similarly, previous research has discussed how women’s participation can lead to empowering experiences and skill-building at the individual level, all of which could be beneficial long-term. The second competing finding is that any post- conflict gendered change is fleeting, and instead, patriarchal backlashes after conflict often reinforce traditional gendered ideas, limiting women’s gains. Relatedly, scholars have found that former rebel women also experience backlash after conflict for their violent deviation from traditional gender roles. Chapters 2 and 3 speak specifically to these competing ex- pectations, finding that women’s participation in rebel groups during conflict can lead to long-term gendered change for women, as groups with legacies of women’s inclusion are far more likely to elect women after war. Additionally, I find that there is far more variation in the post-conflict fates of former rebel women than the literature has previously estimated. While it is still true that many former rebel women are marginalized after conflict, I find that a significant number of these women do successfully move into electoral politics after war, suggesting that they can build on their wartime experiences and skills. Women’s par- ticipation in rebel groups, thus, can lead to gains for women generally, as well as for the specific women participating. Finally, this dissertation demanded the collection of three unique datasets. In the first chapter, I introduce a dataset on women’s party-level political representation in Africa from 1970-2020. The dataset details women’s candidacy and election. This dataset was useful for this study, as it revealed the differences in women’s representation among rebel parties and other parties. However, it is also may serve many uses beyond the scope of this dissertation, as no dataset to my knowledge previously has offered party-level data on women’s representation in Africa. In the second chapter, I present unique party-level data on women’s political representation in a global sample of former rebel parties from 1970-2020. This data offers an understanding of global patterns of women’s inclusion in rebel parties. Finally, in the third chapter, I introduce a dataset on the election of former rebel women 6 in a global sample of rebel parties from 1970-2020. This dataset is the first to offer an understanding of the prevalence of rebel women’s political lives after conflict and could be useful to scholars in the future to understand in greater depth the causes and consequences of the political inclusion (or exclusion) of former rebel women. 7 CHAPTER 1 WOMEN’S POLITICAL REPRESENTATION IN AFRICAN REBEL PARTIES 1.1 Introduction A short two months after coming to power in 1986, Yoweri Museveni, the former rebel leader turned president, delivered a speech on International Women’s Day regarding the status of women in Uganda. He said, “Here is an urgent need to destroy the prevailing (self) defeatist mentality...that they [women] are less capable than men. This mentality is a result of centuries of intimidation and indoctrination and subjugation by men” (Quoted in Boyd (1989): 108). This pro-women statement was followed by a call for the empowerment of women. Museveni’s speech set the stage for decades of symbolic commitments to women. Since coming to power, the former rebel leader has facilitated the political ascension of women, with the appointment of women to several top leadership positions, including vice president and many cabinet positions (Tripp 2000, 2015). Additionally, there has been a significant increase in women’s presence in Parliament. These post-conflict improvements have had relevant effects on perceptions of the state and the level of support for the National Resistance Movement (NRM), the political party born out of the rebellion. The high level of women’s representation has allowed the NRM to expand their stronghold of support among voters (Muriaas and Wang 2012) and has lowered perceptions of corruption (Tripp 2001), with President Museveni even arguing that women have “stabilized politics” in Uganda.1 Similar trends have been seen in Rwanda. Over the past few decades Rwanda has 1 Simmons, Ann and Robin Wright. “Gender Quotas Puts Uganda in Role of Rights Pioneer.” Los Angeles Times, February 23, 2000. 8 transformed from a state ripped apart by war to the country with the highest political representation of women in the world. With women accounting for about sixty percent of representatives in their legislature, Rwanda has become a global leader for political equality and a model for post-conflict reconciliation, with significant praise given to President Paul Kagame, the former rebel leader (Berry 2018a; Mann and Berry 2016). As the Rwandan Patriotic Front (RPF), Kagame’s rebel group, transitioned into a political party, they placed women at the forefront of political reconciliation (Berry 2015b, 2018a). Women politicians were “heralded by the RPF leadership—and more critically, heralded by themselves—as peacemakers, mothers, and as the ‘heart of the home’” (Berry 2015b, 145). Women’s presence in politics, and the narratives that surrounded them, ultimately won the RPF significant levels of domestic and international support (Mann and Berry 2016). The need for former rebel groups that have transitioned into political parties (hereafter “rebel parties”) to signal stability and build support is not an anomaly. Rather, after conflict, these newly formed political parties have a particular need to expand their national support and communicate a change to their identity. However, to my knowledge the role that women play in these strategies has not been studied in the aggregate. This is particularly surprising, given the high levels of women’s political representation in post-conflict countries (Hughes 2009; Hughes and Tripp 2015). In this paper, I ask how women’s political representation after conflict differs in rebel parties. I build an argument that rebel parties will have greater incentives to run women candidates after conflict. Rebel parties face unique challenges after conflict, in that they must quickly build broad support in order to be electorally successful and that they must divorce themselves from violent conflict legacies (Manning 2007; Wittig 2016). I argue that running female candidates allows rebel parties to signal a break from the status quo and to manipulate feminine stereotypes regarding peacefulness. Likewise, this paper argues that rebel parties will be more likely to elect women, as well. Female candidates for former rebel parties will be highly appealing to voters, as they also signal a break from the status quo and benefit from assumptions regarding women’s trustworthiness 9 and accountability. Likewise, former rebel parties and allied women activists will be likely to throw support behind these candidates, increasing their chances for success. This paper presents unique data on the level of women’s political representation at the party-level in Africa from 1970 to 2020. The results suggest that rebel parties run and elect more female candidates than other parties. This paper provides a novel contribution by theorizing how these strategies regard- ing women’s political representation may differ at the political party-level and shows that they are heavily influenced by conflict legacies. Other explanations of post-conflict increases in women’s political representation focus on changing institutions, like gender quotas, or changes to gender roles in society (Hughes 2009; Hughes and Tripp 2015; Tripp 2015). This paper looks beyond aggregate influences to understand how these changes differ at the po- litical party level. Additionally, I couch this argument within the literature on former rebel groups transitioning into political parties. The relationship between gender and rebel party politics has been largely understated in the literature thus far, with no scholarship beyond case studies explicitly considering the role of women in these parties to my knowledge. Such an absence in our understanding is glaring considering the high levels of women’s participa- tion in rebel groups and the evident continuation of women’s inclusion. As such, this article offers an understanding of the nuanced and important role of women in rebel parties. 1.2 Women’s Representation in Post-Conflict States Scholars have consistently found evidence that on average women’s political repre- sentation is higher in post-conflict states (Hughes 2009; Hughes and Tripp 2015; Viterna, Fallon, and Beckfield 2008). Civil conflict creates substantial change at the cultural, social, and institutional levels that allow greater opportunities for women to move into the politi- cal sphere (Hughes 2009). At the cultural level, gendered roles change significantly during conflict. Women step into a number of non-traditional roles within their families and soci- ety that empower them to continue taking on non-traditional roles post-conflict (Bop 2001; 10 Handrahan 2004; Tripp 2015). At the social level, women activists can mobilize at critical points during the conflict resolution in order to ensure that their interests are put on the agenda (Tripp 2016, 2015). Finally, at the institutional level, pressure from international actors influences the implementation of new political institutions, such as gender quotas, that enable increases in women’s political representation (Tripp 2015). These mechanisms create new norms regarding women’s place in political spaces, not only providing women with the empowerment to seek new roles, but also offering greater access to the resources and networks that women rely on to successfully contend for political office (Hughes 2009).2 The subsequent increase in women’s political representation provides significant ben- efits to women, such as the passing of legislation that is relevant to women’s policy priorities (Mechkova and Carlitz 2020) and an increase in levels of women’s civic engagement (Barnes and Burchard 2013). Additionally, scholars have found that greater representation of women is often associated with broad social progress, like greater spending on social welfare, edu- cation, health, and children (Bolzendahl and Brooks 2007; Carroll 2001; Kittilson 2008; Lu and Breuning 2014). Despite the benefits associated with women’s political representation, women’s path to political office is littered with hindrances. Women frequently face negative bias from voters, as they are thought not to hold the characteristics that voters associate with political leaders (Dolan 2010). In particular, Tickner (1999) argues that the associa- tion of women with peace keeps women out of positions of political power, maintaining that gendered assumptions that cast “men as active, women as passive; men as agents, women as victims; men as rational, women as emotional” will limit women’s credibility regarding issues of domestic and international security (4). However, scholars have noted that these associations can actually benefit both women 2 Scholars have aptly described how these advances for women during conflict may diminish after conflict (Bop 2001; Handrahan 2004; Moran 2010). In many cases, patriarchal structures are reinforced after conflict, limiting long-term gains for women (Pankhurst 2003, 2012). This contradiction to political gains by women in the post-conflict environment renders it of even greater importance to understand the mechanisms that lead to these changes. 11 candidates and their respective political parties in certain cases. In particular, Valdini (2019) argues that women in politics are “thought of as more honest, different from the norm, and as an overall symbol of democracy” (42). First, women are often seen as having more “com- munal” characteristics (Alexander and Andersen 1993; Huddy and Terkildsen 1993; San- bonmatsu 2002). Voters express higher perceptions of compassion, empathy, honesty, and trustworthiness among female politicians (Alexander and Andersen 1993; Dolan 2010; Huddy and Terkildsen 1993). Second, women politicians are seen as a break from the norm or sta- tus quo. For example, in contexts in which the public assumes corruption to be widespread, women are seen as independent from these forces (Barnes and Beaulieu 2014; Goetz 2009; Valdini 2013). In countries such as Peru, Mexico, and Uganda, women’s presence was suc- cessfully used in law enforcement and the treasury sectors in order to correct for perceptions of corruption (Goetz 2009). Similarly, Valdini (2019) finds that after political scandals, po- litical parties increasingly promote female candidates in order to distance themselves from corruption and to benefit from perceptions of women’s trustworthiness and honesty. Fi- nally, women’s presence can provide signals about the quality of democracy. Higher levels of women’s political inclusion tend to be associated with democracy, as the population is more equally represented (Valdini 2019). Schwindt-Bayer and Alles (2018) find that higher levels of women’s representation in the legislature is associated with greater support for democracy and higher levels trust in legislators. Similarly, Clayton, O’Brien, and Piscopo (2019) find that women’s presence in decision-making bodies legitimizes perceptions of the institution and the fairness of its practices. By electing more women candidates, political parties can “feminize” their organization and ultimately benefit from the stereotypes discussed (Valdini 2019). These stereotypes are used as heuristic shortcuts for voters, who may apply assumptions about individual candi- dates onto the entire party. Notably, Valdini (2019) finds that these strategies are frequently used after instances of political scandal and declining democratic legitimacy. Contexts of civil conflict mirror such conditions, in which doubt has been cast onto the abilities of previ- 12 ous leaders and as a result, voters seek commitments to peace and stability (Shair-Rosenfield and Wood 2017). In the next sections, I discuss the incentives for political parties in post- conflict environments to strategically increase women’s political representation and how these incentives are tied to their conflict legacies. 1.3 Variation in Conflict Legacies in Post-Conflict Politics The post-conflict environment in particular is defined by an unstable and uncertain balance of power. In some cases, the ‘winning’ actors may have ascended to positions of political power. In other cases, power-sharing agreements may ensure that political power is balanced across actors, resulting in a later struggle for a majority of political power. Re- gardless of the conflict outcome, political parties will inevitably find themselves competing with one another for support and will need a recognizable outward identity and broad pub- lic outreach. However, parties that were associated with the conflict will have additional challenges to overcome. In particular, former rebel groups that have transitioned to political parties will have significant challenges regarding their conflict legacy (Manning 2004, 2007). Approximately 50 percent of rebel groups that engage in civil conflict transform into political parties following the conclusion of war (Manning and Smith 2019; Matanock 2017).3 Further, these parties tend to participate and win seats consistently. Nearly 65 percent of groups that transform into political parties and run for office will continue to regularly participate in elections (Manning and Smith 2019). In 90 percent of elections that former rebel parties have participated in, they have won at least one seat (Manning and Smith 2019). In many cases, rebel parties have been successful in securing a majority of legislative seats (Söderberg Kovacs and Hatz 2016). Rebel parties have two key challenges that they must overcome when they enter post- conflict party politics. First, rebel parties must address their legacy of violence (Wittig 3 Generally, this transformation and participation is most common in Africa (Söderberg Kovacs and Hatz 2016). 13 2016). While conflict legacies also follow non-rebel political parties that held power before or during the conflict, rebel parties will have a greater need to distance themselves from conflict legacies. Rebel parties in particular will need to establish their willingness to “play by the rules” and adhere to norms that they flouted during the conflict. Further, they must demonstrate that they have abandoned the use of violent tactics (Tull and Mehler 2005). Second, rebel parties must significantly broaden their base of support and appeal to mass audiences in order to ensure electoral success (Manning 2007). Rebel parties are tasked with recruiting a large number of candidates, without the established ties that other parties have. As such, rebel parties have to identify strategies through which they can gain this support, while ensuring they meet the ideological demands of their former support base, including former rebels (Sindre 2019, 2016a,b). These needs subsequently influence how rebel parties recruit candidates in post-conflict elections (Ishiyama and Marshall 2015). 1.4 Strategic Benefits of Increasing Women’s Political Represen- tation Eckstein (1965) defines an internal war by actors’ “use of violence to achieve purposes which could be achieved without violence” (133). During conflict, rebels opt to use violence in order to achieve their political goals, rather than using established processes. In the post- conflict period, these actors then must convince voters that they have abandoned such violent strategies and will only use established political processes to achieve goals (Manning 2004, 2007). The need for rebel parties to divorce themselves from legacies of violence majorly distinguishes them from other political parties. All parties have incentives to increase their support, however, rebel parties must build support in a way that also enables them to recreate their image to be one that stresses their dedication to peace. By running more women candidates, I argue that rebel parties can benefit from women’s legitimacy and thus distance themselves from their violent legacies. First, rebel parties can benefit from the feminine stereotypes that follow women can- 14 didates. By conferring gendered stereotypes onto the party, rebel parties can pivot from the association with violence to assumptions of transparency and peace. Given that the public tends to assume that women politicians are more compassionate, empathetic, hon- est (Alexander and Andersen 1993; Dolan 2010; Huddy and Terkildsen 1993; Sanbonmatsu 2002), and dedicated to upholding democratic institutions (Barnes and Beaulieu 2014; Goetz 2009; Valdini 2019, 2013), rebel parties will likely benefit from women’s presence and can explicitly manipulate these stereotypes to serve their needs. Tripp (2001) argues that after the war in Uganda, women were seen as “being outside the mainstream patronage networks and therefore capable of cleaning up excessively corrupt institutions” (117). Further, Presi- dent Museveni argued that the presence of women brought stability and improved the state of their politics because “they tend not to be so opportunistic. They are not so reckless like men.”4 Subsequently, conceptions of women as holding characteristics that are associated with stability and transparency may incentivize party leaders to focus on attracting these candidates, in order to signal a change in tactics. This is particularly the case for rebel parties—as voters will be likely to view women candidates as having the ability to correct issues of corruption or a lack of transparency that prevailed during conflict. Women’s pres- ence in these parties may also serve to convince voters that women are “cleaning up” the organization and will ensure it is more democratic and transparent moving forward. Ideas regarding women’s relationship to promoting peace are especially useful to rebel parties. The essentialist idea that women are more peaceful has been empirically rejected, but the notion remains prevalent in everyday life and has important effects on perceptions of women in politics (Berry 2018a; Tripp 2015). Women are assumed to be far divorced from any violent conflict and are more broadly associated by the public with peace (Sjoberg and Gentry 2007). These gendered assumptions are particularly salient for women who run for political office, with voters assuming that women candidates are more pro-peace than men 4 Simmons, Ann and Robin Wright. “Gender Quotas Puts Uganda in Role of Rights Pioneer.” Los Angeles Times, February 23, 2000. 15 candidates (Berry 2018a; Tripp 2015). Former rebel parties can benefit from the assumption that women have innately peaceful inclinations. By running more female candidates, rebel parties can demonstrate that their new party is dedicated to upholding the young state of peace. Berry (2018a) demonstrates that following conflict in Rwanda, many women’s narra- tives reflected the idea that women were more peaceful (78). Moreover, narratives stressed that men had been responsible for the onset of genocide and for perpetrating the violence, and that it would not have occurred under women’s leadership (Berry 2015a). As these women moved into formal politics in Rwanda, Berry (2018a) notes that they continued to rely on this narrative, as if “strategically using this idea of women as peacemakers to justify their presence” (78). The need to have women in the legislature was stressed not just as bonus, but a necessity, with one MP arguing “Women are peace-actors; they are the ones who carry out peace” (Berry 2015b, 145). The assumption that women have innate peaceful incli- nations offers rebel parties the opportunity to demonstrate that their new party also stands for peace, by putting the narratives of these women at the forefront. This strategy similarly helped Ellen Johnson Sirleaf’s candidacy for president in Liberia in 2005. Despite Johnson Sirleaf’s connection to the former warlord, Charles Taylor, women threw their support be- hind the candidate. They handed out tee-shirts reading, “Men have failed Liberia; let’s try a woman this time” (Massaquoi 2007, 30). Likewise, leading up to the 2001 election in Kosovo, the PDK sought to “overcome the shadow cast on the party by the thuggish behavior of some of its cadres and supporters” during conflict and chose Flora Brovina, a well-known human rights activist and women’s leader, as their presidential candidate (Manning 2007, 262). In each of these cases, these actors with explicit ties to violence benefitted from stereotypes about women’s dedication to peace and the subsequent effects on perceptions of the party. In addition, I expect that rebel parties will be more likely than other political parties to attract a large number of female candidates. Political parties serve as the gatekeepers to politics, often purposefully or inadvertently making it difficult for women to break into 16 the networks needed to run for political office (Sanbonmatsu 2002). In comparison to other parties, rebel parties have a rare need to recruit a large number of candidates, often with- out access to traditional political networks (Ishiyama and Marshall 2015). This provides a window of opportunity for women, as it creates a way for them to more easily break into politics. Aspiring female candidates then may be more likely to run for a rebel party, simply because the network is easier to access. Women’s movements in particular can take advan- tage of these openings. Coalitions of women’s organizations have been especially successful in pressuring governments to increase women’s political representation (Kang and Tripp 2018). Likewise, women’s organizations will often also individually pressure political parties to re- cruit more women candidates (Yoon 2004). These groups often have their own resources and networks through which they can ensure their candidates have the resources and training to be successful. As such, relationships with coalitions of women’s organizations could serve as mutually beneficial for rebel parties and the organizations. Women’s groups can access an opening to run their female candidates, while rebel parties have a supply of candidates with their own resources, networks, and political training. In summary, women’s presence in politics provides a number of important signals to the public, particularly after conflict. Rebel parties will have incentives to run more female candidates. By feminizing the party, leaders can boost their legitimacy by improving perceptions of trustworthiness, honesty, and democracy (Valdini 2019). Most importantly, they can convincingly present a party that is perceived as being highly dedicated to peace— a crucial signal in post-conflict environments. Therefore, by running a higher number of women candidates, rebel parties signal a swift change in identity and can distance themselves from their conflict legacies. Alternatively, rebel parties may be more likely to attract female candidates because there are more openings for candidates. This leads to my first hypothesis: HYPOTHESIS 1: Rebel parties will run higher percentage of women candidates than non-rebel parties. 17 In addition to running more female candidates, I argue that rebel parties will be more likely to see a higher percentage of women elected. While I argue that rebel parties are strategically opting to run more female candidates, I expect slightly different mechanisms to influence the election of women. Unlike women’s candidacy, the election of women is based theoretically on voter choice. Thus, I expect that voters will be more likely to opt to vote for women in former rebel parties, as they represent a break from the status quo. Voters may also see women candidates for rebel parties as holding these parties accountable and thereby increasing stability, increasing voters’ incentive to opt for women in rebel parties compared to other parties. Simultaneously, I expect that rebel parties will be more likely than other parties to throw campaign support behind women candidates, increasing their odds of winning. This ultimately will also ensure that more women are elected to their party. Additionally, rebel parties can benefit from alliances to women’s organizations who want to see women elected. These outcomes allow rebels to both benefit from women’s legitimacy in order to divorce themselves from conflict legacies and to potentially broaden their support, thereby solving the key challenges these parties face. First, female candidates will generally benefit from previously described positive gen- dered evaluations. These gendered stereotypes can help women win elections, particularly in low-information contexts as voters use these stereotypes as heuristic shortcuts (Black and Erickson 2003; Frederick and Streb 2008; Mcdermott 1998). In post-conflict settings where information may be harder to access, these heuristics may provide even more ben- efit to women candidates. Women candidates in post-conflict environments may also be more likely to be successful as well due to the idea that they are challenging the status-quo. Scholars have found that after scandals or crises that damage voters’ confidence in their lead- ership, women candidates are more likely to win elections (Valdini 2019, 2013). Similarly, Brown, Diekman, and Schneider (2011) find that after crises, voters seek candidates who differ from the incumbents, who are seen as partially responsible for the crisis. This implies that women may be seen as more favorable candidates, as in most cases they are unlikely 18 to make up a significant portion of the pre-conflict incumbency. Further, given assumptions that women are less corrupt and more dedicated to democracy (Valdini 2019), voters may have more incentives to vote for women candidates for rebel parties because they believe that they increase party accountability, subsequently offering greater potential stability in the post-conflict environment. Additionally, I expect that rebel parties will have strong incentives to throw their support behind women candidates in order to receive the benefits of their presence, increas- ing the likelihood of women’s electoral success. By working to support the campaigns of more female candidates, rebel parties stand to benefit from the boosts in legitimacy and efficacy conferred onto their party through women’s presence. The ZANU-PF in Zimbabwe attempted to utilize such a strategy leading up to the 2005 elections. The party announced just six weeks before the election that they would field 30 female candidates, twice the num- ber included by the opposition party. The move came after years of activists pressuring for a quota adoption. Yet, the sudden choice left commentators speculating that it was an attempt to capture female votes in the election.5 Despite criticisms over their intention, the party was successful in electing a greater number of women. Rebel parties may also more heavily invest in the campaigns of their female candidates than other parties. In Uganda, the financial resources of the NRM ultimately have played an important role in the electoral success of female candidates (Brechenmacher and Hubbard 2020). As such, fewer women choose to run for opposing parties and are less likely to win when they do (Muriaas and Wang 2012). The benefit of supporting these campaigns will be higher for rebel parties than other parties simply because the payoff of the election of women is higher as it establishes greater trust in the rebel party and decreases assumed ties to violence. Rebel parties can also significantly broaden their support by coopting women’s move- ments and organizations, as these alliances have a powerful influence on women’s electoral 5 “Zimbabwe: ZANU-PF Out Wooing Women.” IRIN News, February 16, 2005. 19 success. Coalitions of women’s groups are active in advocating for improving women’s repre- sentation and in campaigning for women (Kang and Tripp 2018; Yoon 2004). These groups often have their own resources and networks through which they can mobilize to ensure that their women candidates are successful. They even sometimes put their candidates through political training programs (Kang and Tripp 2018). For example, in the 2005 election in Liberia, the women’s movements led a massive campaign to encourage women to register and vote; moreover, they campaigned for women to vote for their preferred candidates. As a result, more women were registered to vote than men in that year and are thought to have had significant influence on election outcomes (Debusscher and De Almagro 2016). Women’s movements therefore can serve as a powerful mobilizing force for voters. By mobilizing and co-opting the support of these groups, rebel parties stand to benefit in the polls. Aligning with these movements provides former rebel parties the opportunity to both increase the likelihood of success for female candidates, as well as benefit from the votes of the women in these coalitions. Through these alliances, women candidates in rebel parties may be likely to be more successful in elections. In Uganda, scholars have noted that President Museveni has used ties to women’s civil society organizations as a strategic tool of cooptation (Goetz 2002; Hughes 2009). Tripp (2001) argues that the NRM strategy has been to “cultivate and appoint high-ranking women leaders to curry favor with women more broadly in order to win their votes and support” (116). This support has significantly expanded the NRM’s stronghold (Tripp 2000). Simi- larly, the ANC in South Africa has successfully co-opted women’s support. The Women’s National Conference (WNC), which brings together a number of women activists and parlia- mentarians, successfully pressured the ANC to adopt a 30 percent voluntary gender quota in 1994. This ensured that more women candidates are included in party lists, but also allowed the ANC to secure the support of the organization, which ultimately helped the ANC bolster its support among women (Waylen 2004). In summary, I argue that women candidates for rebel parties will be more likely to win 20 elections. Female candidates for rebel parties will generally benefit from heuristic shortcuts from voters based on gender stereotypes, as well as ideas that female candidates represent a challenge to the status quo. More importantly, they will reap the benefits of a national party interest in electing women, which will result in more concentrated campaign resources and alliances to powerful domestic constituencies. This leads to my second hypothesis: HYPOTHESIS 2: Rebel parties will elect a higher percentage of women than non-rebel parties. 1.5 Research Design This paper argues that rebel parties will be more likely to run and elect a greater number of female candidates than non-rebel parties, using this as a strategy to legitimize the party’s dedication to peace and stability and to broaden electoral support. To my knowledge, party-level data on women’s political representation has not been systematically collected. Thus, a major contribution of this project is the collection of party-level data on women’s political representation in post-conflict Africa. The sample is restricted to the African region. While this sample may have implications for the generalizability of findings, Africa is an appropriate context to begin understanding the relationship between rebel parties and women’s representation as rebel group to political party transformations have been most common in Africa (Daly 2019; Söderberg Kovacs and Hatz 2016). Additionally, African context includes a number of cases in which rebels parties “won” their conflicts and now hold majority power in government. This allows for an understanding of how conflict specific outcomes influence these strategies, as well as how the consolidation of political power is tied to these strategies. The sample includes all African countries that experienced a rebel group to political party transformation between 1970 and 2020. To identify the sample of countries, I relied on the Uppsala Conflict Data Program (UCDP)’s Armed Conflict Dataset (ACD) (Harbom, 21 Melander, and Wallensteen 2008). I then collected data on the number of candidates fielded and the number of seats received by each party for every post-conflict multi-party election. This data was collected from a variety of sources, including from the African Democracy Encyclopedia Project of the Electoral Institute of Sustainable Democracy in Africa, the African Election Database, and the Inter-Parliamentary Union. 1.5.1 Dependent Variables There are two dependent variables of interest. The first dependent variable, Propor- tion of Women Candidates by Party, is used to test Hypothesis 1, which states that rebel parties will run more female candidates. The measure is calculated by dividing the number of women candidates per party by the total number of party candidates. The second dependent variable is Proportion of Women Elected by Party, which is used to test Hypothesis 2, which expects that rebel parties will elect more women. This variable is calculated by dividing the number of seats received by women per party by the total number of seats received by the party. The data on the number of women candidates and the number of seats held by women were collected from a variety of sources, including the African Democracy Encyclope- dia Project, the Inter-Parliamentary Union, election reports, academic scholarship, and news reports. Table 1.1 displays the average percent of women standing as candidates and the percent of women who hold party seats by party type. On average, 28 percent of candidates in non-rebel parties are women, in comparison to 34 percent of candidates in rebel parties. Among elected MPs, there is a more drastic difference between the two party types, with rebel parties having more than twice the percent of women MPs.6 1.5.2 Explanatory Variables The key explanatory variable is Rebel Party, a dichotomous variable indicating if a political party transformed from a rebel group. Matanock (2016)’s Militant Group Electoral 6 These differences are significant at the 0.05 level. 22 Table 1.1: Average Percent of Women Candidates and MPs by Party Party Type Percent Candidates Non-Rebel Party 28% Rebel Party 34% MPs Non-Rebel Party 8% Rebel Party 18% Participation (MGEP) dataset, which includes a global sample of rebel party transformations between 1970 and 2010, was used to identify these parties. In order to extend the sample to 2020, background research was conducted on each party in elections that occurred between 2010 and 2020 to determine if they were former rebel parties. Approximately 15 percent of the sample includes former rebel parties. As control variables, I include a number of measures relevant to the election of women and party politics. First, I control for democracy. To do so, I use Polity IV data, which operationalizes democracy as ranging from -10 (the most autocratic) to 10 (the most demo- cratic).7 I control for gender quotas, which measures the adoption of gender quotas in a given country. This variable comes from Hughes et al. (2019) and is dichotomous. The adoption of gender quotas has significant effects on the level of women’s political representation, usually leading to broad increases in women’s representation (Hughes and Paxton 2008; O’Brien and Rickne 2014; Paxton, Kunovich, and Hughes 2007a). I additionally include a control variable for party institutionalization, as Wylie (2018) demonstrates that women’s politi- cal representation is higher in parties with higher levels of party institutionalization. This variable comes from the Varieties of Democracy Dataset (Coppedge 2019). In constructing this measures, V-Dem considers the level and depth of organization, links to civil society, relationships to party activists, coherence to party platform, and consistent voting behavior among representatives. It is an interval-level variable ranging from 0 to 1, with 0 referring to the lowest level of party institutionalization and 1 to the highest level. 7 I code the variable as dichotomous, with Polity scores below 6 referencing non-democracies and scores of 6 and above referencing democracies. 23 The variable candidate selection also comes from the Varieties of Democracy (VDem) Dataset (Coppedge 2019). The V-Dem variable is coded as a 5-level ordinal variable which indicates how candidates are chosen, including whether they are selected by the national party or by local leaders. Scholars have argued, however, that it is more useful to consider candidate selection processes as being either inclusive or exclusive, with inclusive processes allowing for local influence or broad based selection by party members and exclusive processes being driven by national party leadership (Hinojosa 2012). For this reason, the variable is recoded as a dichotomous indicator for whether the candidate selection process is driven locally and is decentralized (0) or if it is an exclusive nationalized process (1). This variable is relevant to the analysis because it reflects the degree to which these patterns are driven through nationalized party strategies. Additionally, the model includes a dichotomous variable to indicate whether or not a given party is an opposition party. This variable is based on data from the Varieties of Democracy V-Party Dataset (Lührmann and Al 2020). Accounting for opposition parties ensures that the relationship is not driven by dominant party status, nor that parties with more resources and access to state coffers are more likely to use these strategies and to succeed in electing women. I include two additional country-level control variables. GDP per capita (logged) comes from the World Bank and is intended to capture the level of de- velopment, which is often associated with higher levels of women’s political representation (Viterna, Fallon, and Beckfield 2008). Lastly, I include a variable on women’s civil society participation. This variable comes from V-Dem (Coppedge 2019) and is an interval variable that captures the estimated proportion of women’s participation in civil society organiza- tions. Tripp (2015) demonstrates that women’s civil society mobilization is a key driver of women’s political representation after war. 1.5.3 The Model I use a hierarchical regression model to examine the percent of women candidates and elected representatives in each party. I use a hierarchical model in order to incorporate 24 both party-level and country-level variables in the model. I also include year cubic restricted splines in order to control for any effects of temporal trends on the level of women’s political representation by party and account for any non-linear time trends. The results of this analysis are reported in Table 1.2. 1.6 Results Table 1.2: Multilevel Linear Regression Results % Women Candidates % Women Elected (1) (2) Rebel Party 0.07∗∗ 0.09∗∗∗ (0.03) (0.02) Level of Democracy −0.05 −0.03 (0.10) (0.01) Gender Quotas −0.05 0.03∗ (0.06) (0.02) Exclusive Candidate Selection −0.02 −0.04∗ (0.05) (0.02) Opposition Party −0.02 0.02 (0.03) (0.02) Party Institutionalization 0.09 0.25∗∗∗ (0.23) (0.07) log(GDP per Capita) 0.04 −0.03∗ (0.05) (0.01) Women’s Civil Society Participation 0.14 0.20∗∗∗ (0.29) (0.05) Constant −0.14 0.07 (0.32) (0.08) Country Variance 0.01 0.00 AIC −108.77 −416.69 Log Likelihood 68.39 222.35 No. Obs. 124 811 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Time splines omitted Table 1.2 presents Models 1 and 2. Model 1 tests the effect of rebel party status on the percentage of women candidates a party runs, while model 2 tests the effect of rebel party status on the proportion of women a party elects. Model 1 shows a positive and significant 25 relationship between status as a rebel party and the proportion of women candidates that a party puts forward. Figure 1.1 presents the marginal effects on women’s candidacy by party type. All marginal effects are calculated using the observed value approach (Hanmer and Ozan Kalkan 2013). The figure shows that on average rebel parties are expected to field about 33 percent women candidates, compared to 26 percent in non-rebel parties. These estimates lend support to Hypothesis 1, which argues rebel parties will have a higher proportion of female candidates. 0.20 0.35 E(P roport ion of Women Candidat es) E(P roport ion of Women Elect ed) 0.15 0.30 0.10 0.25 0.05 Non− Rebel Part y Rebel Part y Non− Rebel Part y Rebel Part y Figure 1.1: Estimated Proportion of Figure 1.2: Estimated Proportion of Women Candidates Women Winners Model 1 also shows that democracy and the implementation of gender quotas each have negative, though insignificant effects. While this is counter to conventional thought regarding women’s representation at the national level, these results are consistent with the existing literature and the theory. First, rebel parties have an incentive to present a commitment to democracy, even if their actions threaten democratic stability. As such, fielding women candidates may serve as a distraction method for non-democratic practices (see Berry 2015, 2018). Second, while the use of gender quotas overall should increase the number of women candidates, it creates the necessity for every party to run more women— 26 thereby decreasing the usefulness of this strategy for former rebels. The coefficient for opposition parties and exclusionary candidate selection processes are negative as expected, though insignificant. Party institutionalization, women’s civil society participation, and GDP per capita each have a positive, but insignificant, effect. Model 2 tests the effect of rebel party status on the election of women by party. Based on the results presented in Table 2, there is a significant and positive relationship between rebel party status and the percent of women’s seats. Figure 1.2 presents the marginal effects. The figure shows that non-rebel parties are predicted to have only 8 percent of seats held by women, whereas rebel parties are estimated to have an average of 16 percent women present. This represents a significant increase and lends support to Hypothesis 2. In Model 2, GDP per capita and democracy each have negative effects, though only GDP is significant. The negative coefficient for democracy is again rather unsurprising and echos findings that authoritative governments use women’s political representation as a distraction method (Berry 2018a; Donno and Kreft 2019; Valdini 2019). Similarly, Berry (2015b) argues that this strategy goes hand-in-hand with efforts to secure greater aid. This could help to explain the negative effect of GDP per capita, in that poorer countries seek to increase women’s political representation in hopes of securing more aid down the line. The coefficient for gender quotas is positive and significant. The coefficient for opposition party is negative as expected, but not significant, while the coefficient for exclusive candi- date selection procedures is negative and significant. Finally, party institutionalization and women’s civil society participation each have positive and significant coefficients. 1.6.1 The Effects of Conflict-Specific Contexts Conflict legacies have relevant effects on how former rebels govern in post-conflict environments (Huang 2016). Thus, I test the consistency of these findings under a variety of conflict contexts, including time since conflict resolution, conflict outcome, continuing violence, and women’s participation in rebel groups. First, I consider the effect of time, as it is possible that these strategies are most useful within a certain period of time. Time 27 0.4 0.125 P (P roport ion of Women Elect ed) E(Percent Women Elect ed) 0.3 0.100 0.2 0.075 0.1 No Ongoing Conflict Ongoing Conflict 0 10 20 30 40 Years since Conflict Figure 1.4: Marginal Effect of Ongoing Figure 1.3: Marginal Effect of Time Violence is estimated as the number of years since the end of conflict. This is based on the year of conflict termination listed in the UCDP Conflict Termination Dataset (Kreutz 2010). Models 1-4 in Appendix A show that time does not have a significant effect on the number of women candidates, but does have a positive and significant effect on the number of women elected. Figure 1.3 shows the marginal effect of time on the estimated proportion of women elected. The figure shows modest gains for women within the first ten years. The greatest gains occur between 10 and 25 years after conflict, in which the estimated proportion of women elected doubles. Women’s representation continues to rise steadily with time. While the strategic use of women’s political representation may seem to be of greatest importance in the first 25 years based on these gains, the continual increase decades after conflict is not surprising. A reversal or abandonment of commitment to electing women after so much time spent trying to bolster women’s representation could ultimately result in backlash or a loss of important constituencies. Thus, the payoff of these strategies may be greatest in the first few decades following conflict, while after this the costs of reversal are high. Finally, there is no interactive effect of party type and time on women’s representation. Across time, both parties are increasing their level of women’s representation (though this level is consistently 28 higher in rebel parties), suggesting that perhaps other parties are learning and adapting to mirror tactics of rebel parties, though at lesser levels. I additionally test the effect of conflict outcome on this relationship. I again rely on the UCDP Conflict Termination Dataset (Kreutz 2010) to code conflict outcomes. Mod- els 5-10 in Appendix A test the effect of rebel victory, peace agreement, and low activity outcomes and show that each has no effect on women’s candidacy. Looking to the effect on women’s electoral success, these conflict outcomes consistently have no effect, with the exception of instances in which there was a low activity outcome. The lack of a relationship among the majority of these outcomes reaffirms that the strategic use of women’s political representation is salient across conflict contexts. Rather, rebel parties still must overcome relevant challenges that lead them to look towards women representatives. To better understand how low levels of violence and the threat of continued conflict influences these strategies among rebel parties, I consider instances of ongoing violence. To do so, I test for the effect of ongoing conflict in the country at the time of election. These data come from the UCDP Armed Conflict Dataset (Pet et al. 2002; Pettersson and Öberg 2020). Figure 1.4 shows the negative and significant relationship between ongoing conflict and the election of women by party. The figure shows that the estimated average propor- tion of women elected by party is about 0.12 with no ongoing conflict, while it is just 0.075 with ongoing conflict. There is no interactive effect between rebel party status and ongoing conflict. These findings are consistent with the literature on the effects of ongoing conflict on women’s representation. Hadzic and Tavits (2019) argue that ongoing conflict influences a preference for masculine characteristics associated with security, like aggression and domi- nance, ultimately decreasing the political representation of women. Thus, in instances when the conflict period has not had a definitive conclusion in the mind of citizens or in which it is even ongoing, they may be less likely to buy into feminized appeals by rebel parties. While rebel parties are overall more likely to elect women in the post-conflict environment, this relationship depends on the security and stability of peace in that environment. 29 Finally, I consider how the specific conflict legacies of former rebel groups may in- fluence their inclusion of women after war. I specifically look to their legacy of women’s inclusion. I subset the sample to only include rebel parties and test for the effect of women’s membership during war. This data comes from Thomas and Bond (2015). Model 14 in Appendix A shows that women’s participation in a given rebel group has a positive and significant effect on their post-conflict election of women.8,9 While these findings suggest that women’s participation in rebel groups during war may explain some variation among rebel parties, this does not alter the primary relationships under consideration. Regardless of women’s participation during rebellion, the theory of this paper proposes that rebel parties and non-rebel parties will have different motivations to run and elect women. To ensure that women’s wartime participation is not driving these results, I compare the average percent of women candidates run and elected by rebel parties that did not include women in their organization during war and non-rebel parties and find that rebels that excluded women still consistently elect a higher percentage of women than non-rebel parties.10 1.6.2 Robustness Checks I conduct a number of additional robustness checks. First, I further explore the role of democracy in the theorized relationship. The level of democracy is not only highly related to women’s representation, but it also theoretically is closely tied to the degree to which individual political parties, and rebel parties in particular can manipulate women’s electoral success. I first test to see if there are any interactive effects of democracy and rebel party status on women’s candidacy and election. However, I do not find such a relationship. The 8 Due to a low number of observations, I am unable to test the effect on women’s candidacy. 9 I also consider the effect of wartime ideology, and test whether nationalist ideologies are associated with an increased proportion of women elected. This coefficient on this variable, which comes from the FORGE Dataset (Braithwaite and Cunningham 2020), is positive, but insignificant. It is conceivable that other ideologies, such as leftist or Marxist ideologies, would also have a relevant impact. However, among the sample, there is only a single observation in which a group possessed a leftist ideology, demonstrating the need to look to a global sample to further understand the relationship between rebel group dynamics and their post-conflict influences. 10 These differences are significant at the 0.05 level. 30 effect of democracy remains negative and significant—again suggesting that this strategy is most useful when countries are less democratic and more subject to electoral manipulation. This lends further support to the idea that women’s political representation is used as a political strategy. I also use two alternative operationalizations of the level of democracy. I use V-Dem’s (Coppedge 2019) electoral democracy index and liberal democracy index. Each of these variables are interval variables that measure the extent of electoral competition and integrity and individual and minority rights, respectively. I also split the sample to autocracies and democracies in order to test if rebel parties operate differently in these contexts. Models 15-22 in Table A6 and A7 of Appendix A show that the results remain consistent under these specifications. Second, I consider various operationalizations for gender quotas. As opposed to mea- suring whether a given country has adopted a gender quota, I test whether that quota was actually implemented and whether it was effective in raising women’s political representa- tion. Each of these variables are dichotomous and come from Hughes et al. (2019). As presented in Table A8 of the Appendix, the results remain consistent under these specifica- tions. I additionally consider the role of voluntary party quotas and whether this could have an interactive effect with rebel party status. I find that party quotas have a significant effect on the percentage of women elected, but not the percentage of women candidates.11 Third, I test for the effect of seat share, which is the number of seats won by each party divided by the total number of seats in the legislature. The variable is used to account for patterns in seat share that differ among rebel parties and non-rebel parties. Particularly, in many cases rebel parties have gained a majority of seats in elections. The size and resources available to some of these parties, then, could further their ability to recruit and elect women candidates. 11 Very few parties that use these voluntary quotas (8), and only three of these parties are non-rebel parties. Further, in each of the instances in which non-rebel parties have adopted party quotas, they are major competitors with a rebel party who has also adopted party quotas. Thus, it is possible that among these few cases, the adoption of party quotas could be seen as a further strategy to distance themselves from conflict among rebel parties, while non-rebel parties may opt to adopt these measures to remain competitive. 31 Table A9 in Appendix A shows that the results remain consistent when seat share is consid- ered. Finally, I use an alternate model specification for the second hypothesis, which argues that rebel parties will elect higher percentages of women. Given that a significant portion of the observations of women’s election are zero, this could introduce bias into the model (Wooldridge 2010, 2012).12 To account for this, I use a tobit maximum likelihood estimator that is left-censored at zero. Table A10 in Appendix A demonstrates that the results remain consistent under this specification. 1.7 Conclusion Rebel groups that transition into political parties experience a number of unique chal- lenges after war. They must gain legitimacy as political actors among domestic audiences. Parties are faced with the challenge of signaling that they are willing to follow institutional rules and norms and that they will not opt to use violence to achieve goals (Tull and Mehler 2005). In this paper, I argue that rebel parties can strategically use women’s candidacy and election to meet these challenges. Parties can benefit from gendered stereotypes that follow women candidates. By bolstering women’s inclusion, parties can attempt to feminize the group and how it is perceived. Subsequently, this may lead to higher perceptions of the party as being trustworthy, honest, just, and most importantly, highly dedicated to peace. Addi- tionally, rebel parties face the necessary challenge of broadening their support in order to become not just electorally viable, but to gain enough power to determine political outcomes. By running and electing more female candidates, rebel parties stand to co-opt the supporters of these women, such as women’s movements and organizations. I find that rebel parties do in fact run more female candidates and are more successful in electing them. These findings hold across a number of contexts, including various conflict outcomes. However, the results are sensitive to any ongoing conflict, indicating the limits to which rebel parties can benefit 12 In contrast, very few (fewer than 5 percent) of the observations of women’s candidacy are zero. 32 from “feminizing” their party. The paper also offers preliminary findings that the wartime legacy of women’s par- ticipation may also create variation in the percentage of women that rebel parties elect. I show that rebel parties that included female members during war elect more women than rebel parties that did not include women during war. However, despite this variation, I also demonstrate that on average both rebel parties that included women members during war and rebel parties that did not include women during war elect a higher percentage of women than non-rebel parties, indicating that there is a salient and significant difference between how rebel parties and non-rebel parties relate to women’s political representation after war that is distinct from the wartime legacies of women’s participation. Moreover, these preliminary findings indicate that more work must be done in the future to understand how women’s engagement in war is tied to their wartime inclusion in rebel parties. Such an analysis demands a full examination of the roles that women took on, as well as what function they served to the group. Recent literature has demonstrated clearly that women’s participation in rebel groups heightens both the capacity of groups (Thomas 2020; Thomas and Bond 2015), while also improving perceptions of legitimacy (Loken 2018; Manekin and Wood 2020; Wood 2019). It is highly conceivable that these benefits created by women are drawn into the post-conflict period, but this examination is outside the scope of the argument and analysis presented here. This paper is the first, to my knowledge, to probe the role of women’s representation in rebel parties. The results indicate not only the presence of a relevant relationship, but that much more research ought to be done to understand the nuances of the role of women in these parties. While the bolstering of women’s political inclusion by these parties may seem like a positive development at face value, more work must be done to understand whether these parties simply benefit from the signals of women’s participation or if they actually empower the voices of women MPs. Previous scholarship on women’s representation in post- conflict regimes often suggests that it is the former (Berry 2015b; Goetz 2002). The use 33 of women’s political representation as a political tool is reason for concern. Scholars have shown how autocratic and repressive governments have used women’s political representa- tion as a method to improve perceptions of legitimacy (Donno and Kreft 2019). Further, as Berry (2015b) argues it can be a “smoke screen” for what is actually happening domestically, particularly when repression is commonly used. Increasing women’s presence will not auto- matically equate to increasing women’s influence, thus the higher levels of women’s presence in rebel parties should not be equated with gendered progress. These findings have implications beyond the effectiveness of women representatives. The relationship under analysis has relevance for broader understandings of peace and sta- bility in post-conflict environments. Many of these actors have used women’s political repre- sentation as not only a method of gaining support, but also as a way of decreasing political competition from opposition and minorities (see Berry and Lake 2020). While this provides former rebel parties with greater political control, it ultimately creates more disgruntled populations with reason to mobilize in contexts within which rebellion has been shown to lead to success. 34 CHAPTER 2 LABORED LEGACIES: THE POST-CONFLICT IMPLICATIONS OF WOMEN’S WARTIME PARTICIPATION 2.1 Introduction Women’s participation in conflict defies traditional gender expectations that deem war to be the business of men. After war, however, there is extreme variation in if and how women continue challenging these gender norms. Following civil conflict in Sierra Leone, former female combatants returned to their communities to find that “primordial ties to their pre-war cultural past” had strengthened, leaving a society that not only expected them to return to their former gender roles, but even “collectively grimaced at women’s empower- ment in the aftermath of conflict,” (Lahai 2015, 144-145). In this case, despite the challenges that female combatants posed to gender norms during war, women as a group experienced little change between pre- and post-war periods. In contrast, in Uganda, women’s roles in the resistance struggle turned into “an effective strategy for their empowerment,” (Byany- ima 1992, 129). In the years following the war, President Museveni, the former rebel leader, frequently cited women’s contributions during war as a motivator for their political inclusion after war (Tamale 1999; Tripp 2000, 2015). Why women can build on the wartime contribu- tions of former rebel women in some cases, but not others, is unclear. When can women’s participation in war evolve from being an exception to gender norms to a more permanent revision to women’s political status? Many scholars have debated the transformative effects of war on gender relations. Some note that while women’s experiences during war can be individually empowering, it does not necessarily lead to sustained change after war (Meintjes, Pillay, and Turshen 2002; 35 Pankhurst 2003, 2012). Rather, women are often faced with a patriarchal backlash after war that fosters a “re-traditionalization” of gender roles (Moran 2010). In these cases, women’s wartime contributions are viewed as a necessary measure under the constraints of war, but offer little long-term change. On the other hand, many scholars aptly point to positive changes for women writ large after war. On average, women’s political representation is significantly higher in post-conflict states (Hughes 2009; Viterna, Fallon, and Beckfield 2008) and women see an expansion of their political, social, and economic rights (Tripp 2015). Scholars have offered several explanations for these improvements, including evolving gender norms, international pressures, institutional changes, and the influence of women’s movements (Hughes 2009; Hughes and Tripp 2015; Tripp 2015). From this perspective, women’s wartime contributions are seen as one of the many challenges to gender norms that influence sustained change. In this paper, I aim to understand when women’s wartime participation leads to long-term change among gender roles. In particular, I look to the election of women after war. Similar to violence, politics is a patriarchal tradition (Ni Aolain and Rooney 2007). Women’s engagement in each demands a challenge to traditional gendered expectations and an acceptance of this defiance by relevant voices of power (Gilmartin 2015). A continuation of women’s inclusion in these non-traditional roles in the post-conflict period would suggest that women’s roles during wartime were not merely an exception—but instead that they were the harbinger of women gaining equal footing in the political world. I examine the effect of women’s combat participation in rebel groups during war on women’s representation in former rebel parties after war. The literature on rebel groups that transition into political parties suggests that wartime characteristics frequently predict party behavior after conflict (Ishiyama and Widmeier 2020; Manning 2007; Zaks 2017). Yet, the influence of gender dynamics, such as women’s participation, has received little attention. Further, patterns of women’s political representation in former rebel parties are not well understood. I propose that the divergence in outcomes of women’s post-war political status 36 is the result of empowerment women gain during war, as well as the strategies used by former rebel actors. I expect that women’s empowerment created through altered gendered roles in war could foster a direct increase in women’s representation after war. However, I also argue that much of these gains can also be explained through women’s specific wartime contributions. Through this lens, women’s inclusion is a strategy that former rebel parties adapt to the post-war environment to continue cultivating benefits. I present an argument that rebel parties will continue to foster women’s engagement after war, particularly when it serves a political end. I look to how women contribute to rebel groups during war to understand if and how groups opt to include women after war. In particular, I expect that groups that utilized women to bolster their support by communicating their ideological commitment and fostering domestic and international legitimacy will elect significantly more women after conflict, as they will strategize to use women’s presence to capture similar benefits after war. Women’s inclusion in rebel parties in the post-war environment, thus, is an adaption of rebels’ wartime strategies. To test these theories, I collected a novel dataset on women’s political representation in rebel parties from 1970-2020. I find that rebel groups that generally include women in combat roles are more likely to elect a higher percentage of women after the war. However, I also find that these effects are conditioned on a number of factors. First, when women’s combat participation is associated with greater civilian support, their participation is related to higher levels of women’s representation. Second, in cases in which women’s combat integration results in boosts to the organization’s international support, groups will also elect a higher proportion of women after war. These findings suggest that women’s wartime participation can evolve into sustained political gains for women, particularly when it serves to also create political gains for their party. This echoes arguments by Berry (2018a) and Valdini (2019) that women’s political inclusion is often appropriated by political actors for their own benefit. These findings also demonstrate the pertinence of looking beyond aggregate trends of women’s political rep- 37 resentation after war. Previous explanations for women’s gains in electoral politics after conflict have focused on individual or macro-level changes that drive gains in women’s po- litical access. This, ironically, overlooks the role of political parties in promoting women’s political representation, despite the fact that political parties serve as important gatekeepers to politics (Lovenduski and Norris 2003; Norris and Lovenduski 1995). Thus, this paper offers an understanding of long-term effects of women’s conflict participation, as well as of how specific political parties prioritize women’s inclusion in the post-conflict environment. Finally, the paper sheds light on the role of gender in former rebel party politics and demon- strates that rebel parties use gender to their advantage in similar ways to which rebel groups manipulate gender during war. 2.2 Rebel Party Politics and Conflict Legacies Political parties serve as brokers for women interested in entering electoral politics. To enter what is traditionally viewed as men’s domain, women must align themselves with these gatekeepers (Lovenduski and Norris 2003; Norris and Lovenduski 1995). Scholars have emphasized the individual and aggregate factors that drive increases in women’s political representation after conflict (e.g. Hughes 2009, Hughes and Tripp 2015, Tripp 2015), but have underemphasized the role of parties in facilitating these developments. After conflict, political parties face unique demands in that they must work to rebuild society and (re)gain the trust of the public. These demands are particularly severe for parties who engaged in conflict as a rebel force (Manning 2007). Regardless of these challenges, former rebel parties are important political actors in post-conflict environments, as their buy-in to an institutionalized political system is a key driver of stability and security (Manning 2007; Matanock 2016). Additionally, their post-conflict engagement is frequent. Approximately half of all rebel groups transition into political parties (Matanock 2017, 2018). Further, these parties participate in and win elections consistently (Manning and Smith 2019). The party development of rebel parties is also unique in that they heavily draw 38 on their conflict legacies to build their party. The post-conflict behavior of rebel parties is highly influenced by their conflict-period characteristics (Zaks 2017). Most pertinently, scholars have found that the post-conflict identity and public engagement of rebel parties are driven by their conflict legacies. This consistency is the result of the relationship between the newly formed party and their wartime supporters. Like all parties, rebel parties have varying groups of constituents that they must appease (Ishiyama and Batta 2011a; Ishiyama and Widmeier 2013). However, unlike other parties, their wartime supporters make up a significant proportion of these constituents (de Zeeuw 2008a). Even as rebel parties adopt their organization to fit the demands shouldered by political parties, including structural and attitudinal changes, they will adhere to particular wartime policies and practices to continue appeasing the expectations of their wartime supporters (de Zeeuw 2008b; van de Goor and de Zeeuw 2008). The greatest consistencies between war and post-war behavior of rebel parties can be seen in how they frame their identity and engage with supporters. To maintain the support of wartime constituents, rebel parties will preserve some level of consistency to political stances and goals that they held during war. Acosta (2014) argues that the initial transformation into a political party is only sensible to rebels if it fits within the goals and image that the group created during war. Though former rebels are now expected to use peaceful and democratic means of creating change, their purpose in mainstream politics is to achieve similar ends that were delineated during conflict (Söderberg Kovacs 2008; Wittig 2016). Scholars have found that rebel parties will opt to continue operating along wartime cleavages after conflict (Ishiyama 2019). In some cases, such as the Revolutionary United Front Party (RUFP) in Sierra Leone, this strategy is used to mobilize marginalized communities of former combatants and to ensure that they access the spoils promised in negotiations (Söderberg Kovacs 2019). Examining former rebel parties in Sri Lanka and Bosnia and Herzegovina, Sindre (2019) finds that while wartime cleavages weaken with time, rebel parties still frequently invoke images of themselves as “defenders” of their group and 39 its interests. Only in instances of weakening political support do rebel parties opt to make significant changes to their identity and image (Ishiyama and Marshall 2017). The wartime governance of rebel groups also has relevant influences on their post- conflict engagement with the public (Allison 2010; Huang 2016; Ishiyama and Widmeier 2020). How rebels interact with civilian supporters during war affects how they maintain relationships after war, as well as their level of electoral success. Groups that successfully mobilize civilians during war continue to do so in similar ways after the war (Ishiyama and Widmeier 2020; Zaks 2017). For example, Berti (2016) finds that within Hamas and Hezbollah, social and political practices that promoted engagement with civilian supporters were highly consistent between conflict and post-conflict periods, especially in cases in which the group relied on grassroots mobilization. Huang (2016) demonstrates that rebel groups that relied on civilian aid during war are more likely to support democracy after conflict, as they already relied on proto-governance structures during conflict. Similarly, Zaks (2017) argues that the mass mobilization structures that were implemented by the FMLN in El Salvador during the guerrilla war enabled the group to mobilize supporters after war, enabling them to successfully transition into a political party and maintain widespread support after the war. As long as these strategies continue cultivating support and thus success for the group, rebel parties will continue to interact with supporters in similar ways between war and post-war periods (Ishiyama and Widmeier 2020). While the literature has demonstrated the many ways in which conflict legacies in- fluence post-conflict behavior, it has not grappled with how gendered strategies may persist into post-conflict environments. This is despite the fact that the factors driving rebel party behavior after war are also highly related to women’s roles in rebel groups during war. Still, scholars have offered little attention to how women’s inclusion in wartime may be related to their political inclusion after conflict, as well as the role of rebel parties in fostering this. In the next section, I build an argument that the conflict legacies of former rebel parties will create unique incentives that facilitate increases to women’s political representation. 40 2.3 Why Rebel Parties Elect Women Though rebel groups may abandon particular conflict-period tactics, such as the use of violence, much of the identity and engagement of these organizations remain intact as they transition into formal political parties. I posit that like other legacies that persist into the post-conflict period, rebel parties will continue to ensure women’s inclusion after conflict, given that they did so during conflict. Further, I argue that women’s political inclusion after conflict will be driven by similar forces that influence their integration during conflict. I expect this effect to be driven by two primary forces. First, I expect that women’s participation in war will directly effect their election after war due to empowerment through changing perceptions about gender roles. Second, I expect that women’s representation in these parties after war will also reflect continued strategic efforts of the organization to capture the legitimizing benefits of women’s presence. These strategies will reflect some of the factors that made women’s inclusion so beneficial to rebels during the war, except now the goal will be to win seats instead of to win wars. These strategic choices are made with the understanding that the benefits of women’s inclusion are greater than the costs. Enabling women to access political institutions and decision-making roles comes at a cost to political gatekeepers, both in that it demands they create room for these new actors and in that party leaders risk losing support from those who do not approve of women in political roles (Valdini 2019). After conflict, rebel groups must weigh the costs and benefits of including women, just as they had to do during conflict. Women’s inclusion in fighting forces creates several benefits, for both the women themselves and for their groups. In particular, through their participation women often build skills and networks, as well as experience empowerment as they have the opportunity to serve in non-traditional roles (Bop 2001; Byanyima 1992; Shekhawat 2015). Further, women find themselves impacting politics in a way that they were previously restricted from (Bop 2001; Yadav 2016). I expect that women’s inclusion in rebel parties after war will similarly 41 be driven by a desire to continue such skill-building and networking, as well as a desire to fight for the political change that they sought to create through war. Moreover, women’s participation within rebellion may serve to empower other women to pursue non-traditional political roles and increase acceptance of women in these roles by relevant gatekeepers. Through their participation, women find themselves, often for the first time, having similar influence as men. On the battlefield, many former female combatants note that they felt a sense of equality. In South Africa, a former female combatant for the armed wing of the ANC, the Umkhonto we Sizwe (MK), noted that “In the army, people respect each other. It’s only in the army that I’ve seen equality practiced ... Because of that—because of what we’ve been doing on the ground—we’ve been recognized as equals” (Cock 1992, 163). The empowerment of female combatants and the recognition of women as equals may broadly demonstrate to women that they also can and should take on such roles in politics.1 In addition to the individual empowerment of women, women’s participation in rebellion serves to alter perceptions about gender roles among other rebels and the general public, potentially decreasing the gatekeeping by male rebel party leaders and increasing the odds that the public will vote for women candidates (Bop 2001). For example, in Uganda, Boyd (1989) notes how women’s participation in the NRA during conflict broke many gendered barriers in the minds of male soldiers, which would later influence strategies by President Museveni to integrate women into politics. Further, the framing of women as the “fighters and liberators” of the country deemed their political integration both during and after war as necessary for the advancement of the state (Bop 2001). This leads to my first hypothesis: HYPOTHESIS 1: Rebel parties that had women combatants will elect a higher percentage of women after war than other 1 However, the experience of women in rebel groups varies significantly. Many women experience forced recruitment and sexual violence within these organizations (Cohen 2013a,b). Even in the MK, where women describe feelings of equality, they were still subject to sexual harassment and assault during their tenure (Cock 1992). 42 rebel parties. Women’s participation during war also creates significant benefits at the group-level, such as increased capacity and legitimacy (Loken 2018; Manekin and Wood 2020; Thomas and Bond 2015; Wood 2019). I expect that when beneficial, rebel parties will seek to continue these gendered strategies. While women’s inclusion offers many strategic benefits to rebel groups, not all of these benefits can be explicitly extended to the post-conflict period. For example, having women in their fighting forces often allow rebel groups to commit more lethal attacks (Thomas 2020). In other cases, women’s inclusion in rebel groups bolsters perceptions of legitimacy, increasing their level of support. The implications of women’s participation in each of these cases offers relevant advantages during wartime, but only the latter is relevant after war as the group no longer needs boosts to their fighting capacity or lethality. In the following sections, I consider three mechanisms that boost contributions by women and offer the possibility of cultivating benefits during and after war: ideology, civilian support, and international support. 2.3.1 Women’s Inclusion and Ideology Ideology plays a key role in determining when and where women are found in rebellion. Wood and Thomas (2017) find that rebel groups with leftist ideologies are more likely to employ women on the frontline of combat. These groups, which draw on Marxist theories, emphasize the ability of revolution and class struggle to liberate oppressed people and tend to call for the overthrow of existing social hierarchies, including gendered hierarchies.2 Groups that have leftist ideologies often rely on their integration of these marginalized populations to demonstrate their commitment to their ideals. For example, in Nepal, the CPN(M) explicitly tied their inclusion of women to their goal of destroying gendered hier- archies. This framing of their movement not only served to boost the recruitment of women 2 Thomas and Bond (2015) similarly find that groups that include a specific gender-positive ideology are more likely to have women participants. 43 to the armed group, but also enabled them to expand support from other groups who also supported this outcome (Yadav 2020). Similarly, in Colombia, Herrera and Porch (2008) argue that the choice of FARC-EP to recruit women into their movement was strategically tied to their pivot towards waging a “people’s war.” The group believed that for civilians to truly believe that they represented “the people,” they needed to reflect the diversity of the people. By recruiting women during their transition into an “army of the people,” FARC was able to “legitimize its revolutionary vision” and “further its strategic aims” (Herrera and Porch 2008, 621). After war, the incentives for former rebels to continue embodying their wartime ide- als do not lessen. Instead, former wartime supporters expect rebel parties to act upon the promises that were made to them during wartime (Sindre 2016a; Söderberg Kovacs 2008). When these promises explicitly called for the inclusion of the marginalized after war, sup- porters will expect to see these outcomes. Yadav (2017) notes that following the war in Nepal, the CPN(M) did elect members of parliament that reflected these promises. In the first election following the comprehensive agreement in 2008, the party filled about a third of its seats with women, many of whom were from marginalized castes (Yadav 2017). In Zimbabwe, ZANU-PF frequently connected the role of women during rebellion to their com- mitment to women’s post-conflict political integration. At the 1980 Copenhagen Conference for the Decade of Women, Robert Mugabe, then prime minister of Zimbabwe and leader of the liberation struggle, stated “we learned through the liberation struggle that success and power are possible when men and women are united as equals” (Lyons 2004, 41). Mugabe later directly connected women’s political liberation and the party’s ideological goals, saying If women are not drawn into the public service, into militia, into political life, women are not torn off their stupefying house and kitchen environment, it will be impossible to build even democracy let alone socialism. (Goredema and Chigora 2011, 28) In cases in which rebel groups included women participants and held an ideology that 44 dictated the inclusion of the marginalized in social and political structures, I expect that women will continue to be integrated into the party after conflict. This leads to my second hypothesis: HYPOTHESIS 2: Rebel parties that had women combatants will elect a higher percentage of women after war than other rebel parties, given they had a leftist ideology during war. 2.3.2 Women’s Inclusion and Domestic Support During civil war, rebel groups frequently rely on the support of civilians for critical resources, such as food, shelter, and even intelligence (Kalyvas 2006). The resources and support offered by civilians ultimately can have an influential impact on conflict outcomes (Hultman and Hultman 2007; Wood 2014). Further, how civilians perceive rebels matters as well, as the more favorable the public is towards rebels, the greater threat the group will appear to pose towards the government (Findley and Young 2007). Given the crucial impacts that civilians have on conflict dynamics, rebels have incentives to foster positive civilian relationships, as long as these relationships do not come at a cost. Scholars have demonstrated that one cheap method for rebel groups to foster positive community relation- ships and perceptions of legitimacy is through women’s engagement. Women’s engagement in rebel groups can soften perceptions of the group, making their causes seem more worthy of support (Manekin and Wood 2020). Their inclusion legitimizes the group’s goals, with the public deeming rebel groups, along with their goals and tactics, as more just and defensible when women are members (Manekin and Wood 2020). To garner these benefits of legitimacy, female fighters are often placed in outward roles and used in propaganda to humanize their movement and its goals to relevant audi- ences (Loken 2018; Manekin and Wood 2020; Sanı́n and Carranza Franco 2017; Wood 2019). Women members are often used in community engagement, in an attempt to confer a femi- nine and friendly identity and thus, improve their popular support (Loken 2018; Sanı́n and 45 Carranza Franco 2017; Wood 2019). In Colombia, FARC-EP relied heavily on women to manage community relations. Women were responsible for managing community disputes and for ensuring that women and children had their needs met (Herrera and Porch 2008). Similarly, Viterna (2014) argues that women FMLN combatants in El Salvador were strate- gically used during wartime to “tug at the heartstrings of civilians,” which ultimately made civilians more likely to share their food and resources with the guerrillas (199). As rebel groups transition into political parties, they are faced with the dilemma of building support (Manning 2007). I expect that women’s integration is one aspect of “wartime bureaucracy” that is salient in both conflict and post-conflict periods. Rebel groups that experience these boosts to their domestic level of support via women’s inclu- sion into fighting forces will likely continue trying to capture these benefits when they have transitioned into a political party after war. In these cases, they will instead rely on female politicians to bolster their level of support. While the exact signal that women members send in war and post-war periods changes, the feminizing mechanism remains. During conflict, women’s presence is used by rebel groups to signal a more just movement that is worthy of support by civilians. After conflict, I expect that women’s inclusion in the party may be used to signal that the party is more committed to institutional norms and peace. Women in politics, similar to women in war, are interpreted differently than men by the public. Gen- erally, women are perceived as being more trustworthy, empathetic, and honest (Alexander and Andersen 1993; Dolan 2010; Huddy and Terkildsen 1993; Sanbonmatsu 2002). Women are also thought to be less corrupt and more transparent, ultimately increasing trust in their overall commitment to democratic principles (Barnes and Beaulieu 2014; Valdini 2013). Fi- nally, women in politics are frequently assumed to be more dedicated to upholding peace (Berry 2018b). These stereotypes create a similar legitimizing function among women party members as to what women offer rebel groups during war. This leads to my hypothesis: HYPOTHESIS 3: Rebel parties that had women combatants will elect a higher percentage of women after war than other 46 rebel parties, given they relied on civilian support during war. 2.3.3 Women’s Inclusion and International Support During civil war, rebel groups also frequently rely on the support of international actors. Rebel groups often receive financial and material support from foreign actors that support the goals of these violent organizations. This crucial support can also influence rebel groups’ likelihood of victory by altering their capabilities (Salehyan, Gleditsch, and Cunningham 2011). Yet, to attract foreign sponsors, rebel groups may need to adhere to particular standards of legitimacy. Previous literature has shown that women’s integration into rebel membership is one strategy through which rebel groups capture foreign support (Loken 2018; Manekin and Wood 2020). In particular, female fighters are used in propaganda and in outward-facing roles that enable groups to cultivate positive perceptions of legitimacy at the international level. Further, Allemang (2021) finds that rebel groups with women participants receive greater external support from international actors. By electing women into national office, rebel parties stand to benefit from similar boosts in perceptions of legitimacy from international actors. During war, rebels frequently emphasize the participation of women, often beyond their actual level of participation, to capture positive perceptions. Women are frequently used in rebel propaganda to emphasize their support of and contributions to the movement (Loken 2018). The FARC-EP, for example, would frequently create photo-ops with female fighters that would be widely disseminated with the group’s propaganda (Herrera and Porch 2008; Sanı́n and Carranza Franco 2017). Similarly, in the armed branch of South Africa’s Africa National Conference, the Umkhonto we Sizwe (MK), the woman militant became a popular propaganda tool for the group, even though few women served in combat (Cock 1992). In Zimbabwe, ZANU propaganda showed women holding a rocket grenade, trading information, and cleaning weapons (Lyons 2004, 162). They also, however, frequently showed women rebels completing feminine tasks, like taking care of the sick, to soften perceptions of 47 the militant movement (Lyons 2004). This virtue-signaling allows rebels to curtail an image of themselves that is more just and legitimate (Allemang 2021). The effectiveness of these strategies was described by a former female fighter in Kashmir, who argued Without women’s participation, how could our movement receive sympathy from national and international quarters? I remember guiding women how to organize the protest movements, how to confront and provoke the security forces, and how to behave when the media was around. Women played their part quite well.... (Shekhawat 2015, 102) Groups that have experienced the benefits of the strategic placement of women in well-publicized positions during conflict are likely to envision the continued benefits they could receive through these strategies after conflict. In El Salvador, former FMLN women continued to serve in similar political positions, even after conflict (Viterna 2013). As they did during conflict, many former female fighters were placed in positions for the party that had an outward focus to ensure that the international community was met with a feminized version of the party. This included liaison positions between the command and other domes- tic and international actors, including civil society leaders and international organizations like the United Nations and the International Red Cross (Viterna 2013, 188). By electing women to parliament after conflict, rebel parties stand to similarly improve perceptions of the group, softening their wartime legacies through feminine stereotypes. Scholars have ar- gued that former rebel leaders, such as Yoweri Museveni in Uganda and Paul Kagame in Rwanda, have often used women’s political representation as a strategy to build support after war among international audiences (Berry 2015b, 2018a; Goetz and Hassim 2005). In Uganda, Museveni has been frequently accused of using increases in women’s representation to capture more support among domestic and international actors (Goetz 2009; Muriaas and Wang 2012). In Rwanda, scholars argue that Kagame has strategically used gains in women’s representation to capture greater amounts of foreign aid (Berry 2015b; Mann and Berry 2016). 48 In instances in which women’s inclusion fostered gains in international support, I expect that rebel parties will continue to cultivate these benefits after war by ensuring that a greater number of women are elected to the party. This leads to my hypothesis that: HYPOTHESIS 4: Rebel parties that had women combatants will elect a higher percentage of women after war than other rebel parties, given they relied on international support during war. 2.4 Research Design This paper argues that former rebel parties will elect a higher percentage of women when they integrated women into their fighting forces during war. I suggest that this re- lationship is conditioned on factors that increase the benefits of women’s inclusion in rebel groups, including ideology, civilian support, and external support. To test these arguments, I introduce a novel party-level dataset on women’s political representation in a global sample of 106 rebel parties. The sample extends from 1970 to 2020 and includes all rebel parties that successfully gained at least one seat in the legislature following civil conflict. To iden- tify the sample, I relied on the Civil War Successor Dataset from Daly (2019). The sample includes 57 countries, with an average of 2 rebel parties operating in each country. The regional dispersion of the sample corresponds with previous findings on where rebel-to-party transformations are most common, with about 52% of observations in Africa, 20% in Asia, 11% in the Middle East and North Africa, 9% in Latin America, and 8% in Europe. The sample also includes variation in conflict outcomes, with approximately 37% ending in peace agreements, 21% in rebel victories, 21% in low activity, and just 5% in ceasefires, according to the UCDP Conflict Termination Dataset.3 3 The other 13% are classified as ending in “other” outcomes by UCDP. 49 On average, the rebel parties in the sample successfully gained seats in three national elections, though some groups successfully participated in up to nine national elections. Ap- proximately 42 percent of the rebel parties in the sample only participated in one election during the time frame. On average the rebel parties in the sample win a high percentage of seats in their legislatures, holding an average of 61% of seats. Among these representatives, about 17% are women. One limitation of these data is that it only includes the percent of women elected and cannot account for the number or percent of women who ran as candi- dates. While data on elected members of parliament (MP) are typically publicly available and at times even included in easily accessible election reports, data on candidacy is often much more difficult to find. However, it is common within the literature on women’s political representation for studies to focus primarily on the percent of women elected, rather than the number of candidates (Hughes 2011; Hughes and Paxton 2008; Paxton, Hughes, and Painter 2010; Paxton and Kunovich 2003; Rosen 2013). 2.4.1 Dependent Variable The dependent variable is the proportion of seats won by women in each party. This variable is calculated by dividing the number of women who were elected by the party by the total number of representatives elected to the party in a given election. This variable comes from a novel dataset on women’s party-level representation in rebel parties, in which I coded the number of seats that women received in rebel parties from 1970-2020. To code this variable, I relied on a variety of sources, included parliamentary websites, the Inter- Parliamentary Union, election reports, academic scholarship, and news reports. 2.4.2 Independent Variables The primary explanatory variable, Female Combatants, is used to test Hypothesis 1, which suggests that rebel groups with female combatants will elect a higher percentage of women after war, and Hypotheses 2-4, which suggest that rebel groups with female com- batants will elect a higher percentage of women after war when they had a leftist ideology, 50 civilian support, and external support, respectively. Female Combatants is a dichotomous variable that indicates whether or not a given rebel party included women as combatants during war. This variable comes from the Women in Armed Rebellion Dataset (WARD) (Wood and Thomas 2017). This variable captures groups that included women in roles such as frontline combat, suicide bombers or assassins, and other auxiliaries or members of any defense forces. I opt to use this operationalization of women’s participation over other measures that describe women’s general membership in rebel groups to better capture the employment of women in roles that are traditionally held by men and place women in positions that set them as equal to men. In groups where women are members, but not combatants, it may be more likely that they are fulfilling gendered roles. In contrast, the theory presented focuses on how and why women employed in non-traditional gender roles during conflict lead to their continued inclusion in non-traditional roles after conflict. The other explanatory variables are Leftist Ideology, Civilian Support, and Interna- tional Support. Leftist Ideology is a dichotomous variable, indicating whether or not a group held a leftist ideology during war. This variable comes from the Foundations of Rebel Group Emergence (FORGE) dataset (Braithwaite and Cunningham 2020). Civilian Support is a di- chotomous variable that indicates whether or not rebel groups consistently relied on civilian aid during war. This aid includes both financial and material resources, such as rebel taxes, food, weapons, and materials. This variable comes from Huang (2016). To operationalize International Support, I rely on a measure of external support to rebel groups from the Non- State Actor Dataset (Cunningham, Gleditsch, and Salehyan 2009). This is a dichotomous variable that indicates whether or not a rebel group received support, material or financial, from the government of another state. As control variables, I include a number of measures relevant to the election of women, as well as to conflict dynamics that influence the election of rebel parties. First, I control for the level of electoral democracy in a given country. I rely on data from the Varieties of Democracy dataset, which operationalizes the level of democracy as ranging from 0 to 51 1. Generally, democracy has been tied to higher levels of women’s political representation (Phillips 1995; King and Mason 2001; Bush 2011). On the other hand, scholars have more recently demonstrated that autocracy is also tied to higher levels of women’s political rep- resentation (Donno and Kreft 2019). A variable indicating whether or not a country has adopted gender quotas is also included. This variable comes from Hughes et al. (2019). Generally, the adoption of gender quotas is associated with significantly higher levels of women’s political representation (Hughes and Paxton 2008; Paxton, Kunovich, and Hughes 2007b). A control variable for GDP per capita (logged) is included from the Maddison Project Database (2018). GDP per capita is used to operationalize the level of development, which is frequently related to higher levels of women’s political representation (Viterna, Fal- lon, and Beckfield 2008). Control variables for female labor force participation and fertility levels are also included. These variables come from the World Bank. Women’s labor force participation is relevant, as it relates to how many women may have relevant career experi- ence that would later lend towards electoral politics. Likewise, fertility levels are expected to reflect the degree to which women are relegated to the home. Each of these variables has similarly been found to have a relevant effect on the recruitment of women into rebel groups (Thomas and Wood 2018). 2.4.3 The Model To understand the relationship between women’s inclusion in rebel groups and women’s representation in rebel parties after war, I utilize a tobit maximum likelihood estimator that is left-censored at zero. A tobit model is appropriate given that approximately a third of the sample is made up of observations in which a given political party elected zero women. Tobit models are useful in overcoming issues that arise when a sizable proportion of observations are clustered at one end of the distribution of possible observable values (Wooldridge 2010, 2012). In some cases, especially among smaller political parties, no women compete for seats in a given party. Thus, treating these cases as equivalent to women running and not winning (i.e. specifying a linear model) would lead to bias. Tobit models overcome this issue of bias 52 by accounting for the fact that a disproportionate size of observations are clustered at the lower bound of observable values (Wooldridge 2012). The unit of analysis is party-year. The standard errors in all models are clustered by country. 2.5 Results Table 2.1 presents the results of the regression analyses. Model 1 presents the bivari- ate tobit analysis, while Models 2-6 present the full tobit analyses. In Model 2, the effect of women’s participation in combat is positive and significant. This offers support for Hy- pothesis 1, which suggests that rebel groups that have female combatants during war will be more likely to elect women to their party after war. Figure 2.1 shows the marginal effect of women’s inclusion as combatants during wartime. I estimate that rebel parties that do not have women will fill approximately 8 percent of seats with women after war. In contrast, I estimate that groups with women combatants will fill about 23 percent of their seats with women, an increase of 187.5 percent. Models 3-6 in Table 1 present the tobit analyses with the interactive relationships of interest. To interpret these interaction terms, I plot the differences in marginal effects of each. Hypothesis 2 posits that rebel groups that included female combatants will be more likely to elect more women after transitioning into a political party, given that they have a leftist ideology. Figure 2.2 shows the estimated difference in the proportion of seats held by women when rebels include female combatants during war and when they do not include female combatants. Interestingly, when groups do not have a leftist ideology, the difference in the proportion of seats held by groups with and without female combatants is greater than it is for groups with leftist ideologies. This does not offer support for Hypothesis 2. Further, it suggests that leftist groups are not delivering on promises of egalitarianism. Though this is not what was theorized, the finding echoes discussions by Henderson and Jeydel (2013) on leftist groups failing to prioritize issues of gender equality following war. Moreover, interestingly, the finding also suggests that non-egalitarian groups offer further 53 Table 2.1: Tobit Model Dependent Variable: % Women’s Seats by Party (1) (2) (3) (4) (5) (6) Female Combatants 0.269∗∗∗ 0.207∗∗∗ 0.226∗∗∗ 0.240∗ 0.163∗ 0.210 (0.055) (0.054) (0.063) (0.098) (0.080) (0.120) Democracy -0.217 -0.239 -0.220 -0.225 -0.242 (0.127) (0.132) (0.125) (0.129) (0.132) log(GDP per Capita) 0.015 0.016 0.020 0.019 0.023 (0.022) (0.024) (0.026) (0.023) (0.026) Gender Quotas 0.173∗∗∗ 0.178∗∗∗ 0.175∗∗∗ 0.177∗∗∗ 0.181∗∗∗ (0.039) (0.042) (0.042) (0.040) (0.044) Fem. Labor Force Part. 0.003∗∗ 0.003∗ 0.003∗ 0.003∗∗ 0.003∗ (0.001) (0.001) (0.001) (0.001) (0.001) Fertility Levels -0.0430∗ -0.044∗ -0.040 -0.041∗ -0.040 (0.018) (0.018) (0.023) (0.018) (0.022) Leftist Ideology 0.064 0.134 0.068 0.071 (0.061) (0.083) (0.065) (0.062) Civilian Support 0.219∗∗∗ 0.224∗∗∗ 0.249∗∗ 0.223∗∗∗ 0.246∗∗ (0.052) (0.051) (0.088) (0.049) (0.091) External Support 0.209∗∗∗ 0.216∗∗∗ 0.209∗∗∗ 0.176∗∗ 0.188∗∗ (0.048) (0.049) (0.048) (0.064) (0.067) Female Combatants* -0.083 -0.064 Leftist Ideology (0.077) (0.082) Female Combatants* -0.039 -0.026 Civilian Support (0.112) (0.115) Female Combatants* 0.064 0.049 External Support (0.098) (0.099) Constant -0.065 -0.402 -0.434 -0.482 -0.447 -0.515 (0.058) (0.260) (0.300) (0.372) (0.265) (0.375) Sigma 0.050∗∗∗ 0.021∗∗∗ 0.021∗∗∗ 0.021∗∗∗ 0.021∗∗∗ 0.021∗∗∗ (0.010) (0.005) (0.005) (0.004) (0.004) (0.004) No. Obs. 211 138 138 138 138 138 Pseudo R 2 0.340 1.905 1.923 1.908 1.916 1.929 Log Likelihood -53.38 26.88 27.41 26.97 27.21 27.60 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 opportunity for women in politics when they included women during war, reinforcing the idea that women’s integration into these organizations is more of a reflection of strategy than 54 it is of principle. Figure 2.1: Marginal Effect of Female Combatants Hypothesis 3 suggests that rebel groups that included female combatants will be more likely to elect more women after transitioning into a political party, given that they rely on civilian support. Figure 2.3 shows the estimated difference in the proportion of seats held by women when rebels include female combatants during war and when they do not include female combatants for groups that did not use civilian support and those that did use civilian support. For groups without any civilian support, those with female combatants are estimated to fill about 9 percentage points more seats than groups without female combatants. In contrast, among groups that rely on civilian aid, those that include female combatants fill about 25 percent of their seats with women compared to groups without female combatants, who fill about 7 percent of their seats with women. Overall, this demonstrates support for Hypothesis 3, though the relationship between civilian support and female combat integration appears to be more additive than interactive. 55 Figure 2.2: First Differences of Female Combatants and Ideology Hypothesis 4 posits that rebel groups that include female combatants will be more likely to elect women after transitioning into a political party, given they also rely on external support. The results suggest that there is a positive and significant relationship between these factors. Figure 2.4 shows the estimated difference in the proportion of women held seats when rebels include female combatants during war and when they do not include female combatants for groups that did not use external support and those that did use external support. Comparing instances of no external support and external support, there is a significant difference between the effect of women’s inclusion only when groups relied on external support, demonstrating an interactive effect. I estimate that groups that rely on external support and have women combatants will fill about 28 percent of their seats with women, compared to filling just 10 percent of their seats with women when they did not include women combatants during war. When external support is not used, the difference in women’s integration is not significant. These findings lend support to Hypothesis 4. 56 Figure 2.3: First Differences of Female Combatants and Civilian Support As for controls, the level of democracy is negative but insignificant. Though unex- pected, the negative effect is not necessarily surprising. Scholars have argued that autocratic governments may use women’s political representation as a political tool to garner more le- gitimacy (Berry 2015b, 2018a; Donno and Kreft 2019). In this paper, I have similarly argued that rebel parties will opt to elect women when it is politically useful. As such, the incen- tives for these parties to include women may also be greater in the light of lower levels of democracy, as the need to benefit from the signal of legitimacy offered by women is even greater. GDP per capita has a positive but insignificant effect. Gender quotas and female labor force participation have a positive and significant effect. The level of fertility has a negative effect, as expected, though insignificant. 2.5.1 Alternative Explanations I consider a few alternative explanations that could drive women’s political represen- tation in rebel parties. Previous theories on women’s political representation after war have 57 Figure 2.4: First Differences of Female Combatants and External Support examined changing gender roles, institutional changes, international pressure, and women’s civil society movements (Hughes and Tripp 2015; Hughes and Paxton 2008; Tripp 2015). Models 1-5 in Table B2 of Appendix B account for these factors. To operationalize chang- ing gender norms, the model includes women’s labor force participation and fertility levels (World Bank). As gender norms change, more women will likely enter the workforce, raising their participation level, and would likely begin having fewer children, decreasing the fertility levels. Women’s labor force participation has a positive and significant effect, while fertility has a negative but insignificant effect. To account for institutional changes, the model in- cludes gender quotas (Hughes et al. 2019). This has a positive and significant effect. I include the net Official Development Aid (logged) received by a country in a given year to opera- tionalize international influences (World Bank). This variable is positive, but insignificant. Finally, I include a variable from V-Dem to account for the level of women’s participation in civil society movements and find that this has a positive but insignificant effect. In summary, 58 while gender quotas and changes to gender roles (as operationalized through women’s labor force participation) may explain some of the variation observed, much of it is due to wartime factors, including women’s combat participation and wartime bureaucracies. These findings suggest the importance of carefully examining the effect of wartime characteristics of rebel groups, as well as the need to look below aggregate-level explanations. In addition, I consider the role of voter perceptions of women’s representation, as it is difficult to disentangle whether the observed relationship is really driven by rebel party strategy as opposed to voter preferences or both. If voter perceptions of women are driving the results, this should theoretically be reflected in the model through variables that measure some degree of women’s status in the country. In Models 1-6 in Table B2 in the Appendix, I test for the effect of two proxies on the level of women’s representation in rebel parties. I include a variable from V-Dem that measures women’s property rights. This is an ordinal variable that captures if no, some, many, most, or all women enjoy property rights in a given country (Coppedge 2019). I use the Exclusion by Gender Index from V-Dem, which is an interval variable that measures the degree to which women are denied access and participation to space in governed places (Coppedge 2019). Neither of these variables have a significant effect on the relationship and the results remain consistent. Finally, in Model 6 in Table B2 in Appendix B, I include country-fixed effects in order to attempt to control for any additional country-level variation in the public’s willingness to vote for women. The results remain consistent under this specification. In summary, none of the variables that capture the public’s perceptions of women or their status in the country have an effect on the observed relationship. Instead, the relationship appears to be driven by conflict-level factors, as theorized. I also consider the role of seat share in the legislature in influencing the level of women’s political representation in each party, as it is possible that larger and more powerful parties simply have more resources to devote to recruiting and electing women. In Models 7 and 8 in Table B3 in Appendix B demonstrate the effect of seat share, measured as the 59 Figure 2.5: Marginal Effect of Seat Share and Female Combatants proportion of seats each party holds in the legislature, on the percentage of women elected to the party. These data on seat share were collected from the Inter-Parliamentary Union. I find that seat share does not have a significant effect on women’s election. However, Model 8 tests for an interactive effect of seat share and women’s combat participation. Figure 2.5 shows that there is a signification effect on the election of women. Parties that did not include women as combatants during war elect a consistent proportion of women after war across seat share. However, parties that did include women during war elect a greater proportion of women as their seat share increases. This suggests that more powerful parties do in fact elect more women, but only when they also have a legacy of women’s inclusion. Lastly, I test for the effect of opposition party status on the election of women, as ruling parties tend to be more successful in electing greater proportions of women. Models 9 and 10 in Table B3 in Appendix B show that opposition party status has a positive and significant effect on the election of women. Figure 2.6 plots the marginal effect of the inclusion of female 60 Figure 2.6: Marginal Effect of Opposition Parties and Female Combatants combatants and opposition party status. The figure demonstrates that while ruling parties tend to elect more women than opposition parties, in both cases parties that included women as combatants during war will elect more women after war, offering support to the primary results. 2.5.2 Robustness Checks I conduct several robustness checks. First, I use a number of alternative explanatory variables. I test for the effect of the proportion of women holding combatants roles during war, as it is reasonable to expect that higher levels of women’s integration may lead to greater gains after war. This variable comes from the Women in Armed Conflict (WARD) dataset (Wood and Thomas 2017). Model 11 in Table B4 of Appendix B demonstrates that a greater proportion of women in combat roles creates positive and significant effects on women’s political representation. I also rely on alternative measures of GDP per capita and democracy. I use World Bank data to operationalize GDP per capita and find that the 61 results remain consistent. I include a variable from Polity2 to operationalize the level of democracy and again find consistent effects. I also consider differing operationalizations of gender quotas, testing whether implemented quotas or quotas that have effectively increased women’s representation have a differing effect. These variables come from Hughes et al. (2019). The results remain consistent. I also use alternative model specifications. Table B5 in Appendix B uses a linear probability model to test the relationships under analysis. Model 16 includes year fixed- effects, while Model 17 includes cubic splines for time. The results remain consistent across these specifications. I also specify the model using a zero-inflated beta regression analysis, to test whether there is a differing relationship between whether or not parties elect any women and how many are elected if so.4 Table B6 presents the zero-inflated model and demonstrates that while there is a consistent relationship between the explanatory variables and the proportion of women elected, there is not a significant relationship to whether parties elect women at any level. This finding does not diverge from the theoretical expectations, as the theory aims to differentiate parties that elect a higher percentage of women, not whether they elect any women at all. Moreover, to better understand why certain parties elect no women, the relationship to how many women candidates are run also ought to be specified, which is currently outside of the scope of the paper. Finally, in Table B7 of Appendix B, I include a Heckman Selection Model that accounts for the role of female combatants in rebel- to-party transformations. I find that while the participation of female combatants has a positive and significant effect on a group’s transition, the hypothesized relationships remain consistent when this is taken into account. I also consider the effect of several conflict period rebel characteristics. In Table B8, I test for the effect of other ideologies including Islamist, nationalist, non-left secular, and religious ideologies. These variables come from the FORGE Dataset (Braithwaite and Cun- 4 I utilize multiple imputation to specify the two-stage model, as due to missing data there are too few observations to otherwise identify the model. 62 ningham 2020). Only Islamist ideologies have a significant effect, decreasing the number of women elected. I also test for the effect of troop size and the presence of UN peacekeepers. These variables come from UCDP and Matanock (2016). Neither troop size nor the presence of UN peacekeepers has a significant effect. In Table B9, I account for rebel governance structures, in order to better understand how this wartime variation influences the relation- ship. In Models 28-34, I consider whether rebels had legal institutions in place, if they held elections, if they had police forces, if they offered humanitarian aid, if they offered educa- tion services, and if they conducted any international diplomacy. Each of these variables is dichotomous, indicating whether a given group had the specific institution or not, and come from the Rebel Governance Dataset (Huang 2016). I find that these variables, with the exception of humanitarian aid, have no significant effect on the percentage of women elected. The results remain consistent under these specifications, regardless. I next test for the effect of the number of years since the end of conflict on the pro- portion of women elected, to understand if these strategies by rebel parties are most salient immediately after conflict or if they are employed long-term. I find that time has a positive and significant effect on the percent of women in rebel parties. Figure B1 in Appendix B shows the marginal effect of the interactive relationship between female combatants and time and shows that over time both rebel parties that had female combatants and rebel parties that did not have female combatants elect a greater proportion of women, though the pro- portion is consistently higher for rebel parties that included women. Lastly, I consider the effect of conflict outcome on the theorized relationship. To do so, I use data from the UCDP Conflict Termination Dataset. Table B11 in the Appendix presents the results. I find that rebel victory has a positive and significant effect. I test for an interactive effect between rebel victory and the inclusion of female combatants. Figure B2 presents the marginal effects of this interactive relationship and shows that parties that included female combatants during war and were victorious in their conflict elect a significantly higher proportion of women. The coefficient for conflicts that end in peace agreements is positive, but insignificant. Con- 63 flict terminations that included either cease fires or a low activity outcome have a negative and significant effect on the proportion of women included in rebel parties. The primary relationship between female combatants and the percent of women in rebel parties remains consistent, however. 2.6 Conclusion Rebel groups that transition into political parties are affected by their conflict legacies in a myriad of ways. It influences the policies they prioritize (Acosta 2014), the candidates that they recruit (Ishiyama and Marshall 2015; Sindre 2016a), how they interact with their constituency (Manning 2007), and their commitment is to democracy and rule of law (Huang 2016). In this paper, I show that the conflict legacies of rebel groups also influence their post-conflict election of women. I argue that rebel groups that had female combatants during conflict will be more likely to elect women after conflict, particularly when they can draw on similar benefits of women’s participation after war. I expect that in instances in which groups relied on women’s participation to communicate their ideological commitment and bolster their level of support, via civilian aid and external support, they will be more likely to elect women after war because there are opportunities for women to continue creating similar boosts in their level of support. In essence, rebel groups who experienced the benefits of women’s labor during war will be more likely to continue seeking the fruits of this labor after conflict. Women’s integration into these organizations is a political strategy that allows actors to cultivate greater legitimacy—a strategy that is useful both during and after war. I find that while women’s combat participation alone does lead to increases in the percentage of women elected to the party after war, the degree of these gains also depends on how women served in rebel groups during war. Moreover, it is related to two important wartime legacies that also influence the post-conflict behavior of rebel groups–domestic and international support. I find that groups that relied on civilian and international aid during war and included female combatants are more likely to elect a higher proportion of women to 64 their party after war. Though the party’s goals may have changed between war to post-war periods, these groups may still seek to benefit from the boost to domestic and international legitimacy that they experience from women’s inclusion. Further, continued support from these actors may be predicated on women’s continued integration. The results do not reflect the theorized interactive relationship between leftist ideology and women’s combat roles. I find that women’s inclusion offers greater gains in the number of seats that women fill in parties with non-leftist ideologies than for those with leftist ideologies. This not only indicates a failure of leftist groups to act upon promises of egalitarianism, but it demonstrates that non-egalitarian groups are using these gendered strategies at great levels, suggesting that women’s integration is more about political benefits than ideological ideals. Finally, I compare these explanations to previous theories on women’s representation after war, and find that wartime factors offer stronger links to women’s party-level representation. These findings have significant implications for understandings of former rebel party politics, as well as for the political representation of women after conflict. Gender has been relatively understudied in the field of rebel party politics, to my knowledge. Thus, this study offers a new understanding of how former rebel parties engage with women after conflict and how this is related to their wartime policies. The results suggest that the labor of women during war has long-term positive effects on the integration of women into politics. Women’s roles in rebel groups can serve as an opportunity for political engagement in contexts where such political access is rare for women. These results suggest that the impacts of their labor are not limited to the conflict period, but rather that it continues to create opportunity for women after conflict. This study cannot speak to which women these post-conflict political opportunities are awarded to. In future research, such patterns should be considered. Scholars have demonstrated that rebel groups frequently turn to their former combatants to recruit party members after war (Sindre 2016a,b). Given the high levels of women’s inclusion in these rebel parties after war, it is conceivable that they may turn to the same women that they 65 relied on during war. Relatedly, this paper does not consider candidate recruitment. The scope of the theory is limited to the incentives for rebel parties to ensure that women make it into their seats. This does not include how they select individual women, nor how women fare in elections. Given the unique conflict legacies of rebel parties, it is likely that candidate recruitment is highly conditioned on the wartime experiences of women in the country gen- erally and within their rebel organization. This nuanced issue ought to be fully considered within future work. Finally, this paper speaks to the representation of women after war. Though scholars have shown that women’s political representation increases significantly af- ter conflict, the influence of political parties in this trend has not been considered, to my knowledge. This paper demonstrates that parties that have histories of women’s inclusion may be the ones who prioritize it after conflict as well. 66 CHAPTER 3 THE ELECTION OF FORMER REBEL WOMEN 3.1 Introduction After conflict, former female rebels often face severe discrimination and marginaliza- tion (Coulter 2008; Denov 2008; Doerrer et al. 2011; Henshaw 2020). Their distinct and dramatic deviation from traditional gender roles to use violence is seen as “breaking sacred laws;” as a result, they are considered “impure” (Bop 2001, 29). Rebel women returning home often find that their community members now fear them due to their roles in the con- flict and they are frequently ignored or even ridiculed and provoked (Burman and McKay 2007). These social sanctions have reverberating consequences in other areas of former rebel women’s lives. Research has frequently found that after conflict these women have little eco- nomic opportunities and often live in poverty (Bop 2001; Burman and McKay 2007; Coulter 2008; McKay 2005). Scholars have concluded that these marginalizing forces also extend to the political lives of female former rebels, reporting that these women often find themselves pushed back into traditional roles after war (Barth 2002; Farr 2002). These experiences differ drastically from those of men. Male ex-combatants generally are better able to reintegrate into society (Annan et al. 2011). Moreover, male former rebels frequently find opportunities to move into politics after war, particularly in instances in which their former rebel groups transition into political parties (Ishiyama and Marshall 2015; Manning 2007; Sindre 2016b). Do former rebel parties also create these opportunities for their female former members? Select narratives of former rebel women entering politics after conflict suggest that it is possible for some women to move beyond this marginalization into political positions. During the Ugandan civil war of 1981-1986, Janet Mukwaya served as a captain within the National 67 Resistance Army (NRA). Upon the NRA’s victory, Yoweri Museveni, the former rebel leader who had just ascended to the presidency, appointed Mukwaya as Minister of Gender, Labour, and Social Development (Goetz 2002; Tripp 2015). In 2006, Mukwaya successfully ran for parliament, representing the former rebel party. Likewise, in Nepal, Hisila Yami had a similar path to political power. After several years working as an engineer, Yami left her job to go underground with the CPN(M) and served as a central leader in the movement (IDEA 2011). In 2008, Yami was elected to the Constituent Assembly in Nepal, representing her former rebel group. Mukwaya and Yami have each cited their wartime roles as an important catalyst for their political careers, arguing that they gave them the skills, connections, and motivation to pursue politics (Commonwealth Secretariat 1999; IDEA 2011). The political trajectories of Mukwaya and Yami, though untraditional compared to the average member of parliament, are rather standard among former rebel groups that transition into political parties. Rebel groups frequently recruit former members as they build political parties after war (Ishiyama and Marshall 2015; Sindre 2016b). In Uganda, Mukwaya is one of many former NRA fighters who has entered electoral politics (Tamale 1999; Tripp 2000, 2015). Similarly, in Nepal, the CPN(M) elected many of their former cadres after the war (Ishiyama and Marshall 2015). Based on the literature on the post-conflict marginalization of former rebel women, the political trajectory of Mukwaya and Yami may be an anomaly among former rebel women. However, it is possible that Mukwaya and Yami, like other former rebels, were successful in overcoming post-conflict marginalization due to the help of their former rebel organization. This paper seeks to understand if and how former rebel parties mitigate the marginal- ization of former female rebels after war. When and why do rebel parties elect female ex-rebels? Or, are Mukwaya and Yami truly an exception to the norm of post-conflict marginalization of former female rebels? I argue that women’s contributions during war are the primary factors that influence the post-conflict political integration of former rebel women. These contributions influence party-level attitudes on women’s political inclusion, 68 as well as affect how the party frames the inclusion of these women to the public. However, these changing attitudes about women’s leadership may also influence the general inclusion of non-rebel women to the party, which could likewise serve to decrease the proportion of women in the party who are former rebel women. To test these arguments, I introduce a novel dataset on the election of former female rebel members to former rebel parties. The dataset documents the number of former female rebels elected to former rebel groups that transitioned into political parties from 1970 to 2020 in a global sample of post-conflict envi- ronments. I find that women’s roles during conflict influence their post-conflict integration into rebel parties. In groups where women served as combatants, they are more likely to be elected after conflict. However, I also find that as there is a greater supply of qualified women in a given context, a lower percentage of former female rebels are elected, suggesting that as parties attract more women it may come at a cost to their female former members. This paper is the first to my knowledge to examine the post-conflict political lives of former rebel women using a large-N analysis. Important works by Viterna (2013), Mugabme (2000), Yadav (2016), Giri (2021) and others have offered critical insight into the post- conflict political experiences of former female rebels through qualitative case studies. I build on the careful findings of these scholars to develop expectations about where the post-conflict political opportunities of these women are greatest and where these women are most likely to be politically marginalized. Though large-N analyses cannot capture many of the important individual-level factors that drive women’s entrance, or their exclusion, understanding the political role of former female rebels at the group level is an important step in uncovering the vast divergence in rebel women’s post-conflict experiences. Moreover, understanding the political inclusion of former rebel women is of relevance to descriptive representation in post-conflict legislatures. If political mobilization is carried out by organizations that systematically exclude women—specifically former rebel women—then women’s descriptive representation will likely be worse. Former rebel women may see few avenues to channel their political interests and concerns addressed in formal politics. 69 The post-conflict marginalization of former rebel women is also not trivial to post- conflict peace duration and security. How former rebels are generally reintegrated is thought to be critical for ensuring a stable peace after conflict (Hartzell and Hoddie 2003). In particular, when former combatants are integrated into government and have the ability to influence political change, there is generally a longer duration of peace (Manning 2007; Marshall and Ishiyama 2016; Matanock 2016). However, such efforts to promote peace are likely to fall short if former rebel women are not integrated into such pathways to power. Given that former female rebels face stronger marginalization (Burman and McKay 2007; Coulter 2008; Denov 2008), avenues for their political inclusion may be even more important for protecting peace prospects. An exclusion of former rebels from politics and society more broadly may give certain actors reason to believe their lives were better in some ways during conflict and creates a motivation to return to that kind of environment, threatening the state of peace. Thus, the exclusion of former rebel women from pathways to power is significant to our understanding of post-conflict politics, but also to peace and security. 3.2 The Election of Former Rebels In post-conflict environments, former rebel parties play a critical role in politics. Scholars have found consistent evidence that the inclusion of these former violent actors in electoral politics ensures that they can influence political decisions and thus have few incen- tives to reignite conflict (Ishiyama and Batta 2011b; Matanock 2017). Moreover, former rebel parties participate in electoral politics frequently in post-conflict states. Nearly fifty percent of rebel groups transition into political parties (Manning and Smith 2019; Matanock 2016). Among these parties, most consistently participate in and win seats in elections (Manning and Smith 2019). Moreover, in many cases, former rebel parties have held a majority of 70 seats in parliament, indicating a significant political influence among these parties.1 These political parties are frequently led by and composed of former members of their respective rebel groups (Manning 2007; Sindre 2016a,b). While rebel parties recruit new members to run for party seats, former combatants create a particularly important foundation for the party (Ishiyama and Marshall 2015). These former members have already demonstrated their ideological commitment and loyalty to the organization and its goals, offering a stable base for the party (Sindre 2016b). This ensures that the party has represen- tatives who it can trust. Additionally, these former members are often particularly effective at mobilizing voters (Sindre 2016a). Moreover, rebel parties must be cognizant of the cost of excluding former members. Sindre (2016b) argues that when rebel groups transition into political parties, former members expect that they will be integrated into the party as a form of spoils for their wartime participation. Were the party to exclude those whom they relied on to win from the spoils of victory, it could lead to a decrease in the party’s support base, but more importantly, could threaten peace and stability (Manning 2007; Söderström 2013). Among their former members, rebel parties are discriminatory in whom they recruit as candidates, focusing on the most electable members. As Turshen (2002) notes, for many former members of rebel groups, “their skill was fighting, not politics,” (73). Among former members, then, the potential of each individual is not considered to be equal. Instead, rebel parties often turn to their former members with relevant skills and experience to run for party seats (Ishiyama and Marshall 2015; Manning 2007; Sindre 2016b). These members are those who are politically competent, capable, and educated (Manning 2007). This most frequently includes elite members, who served in roles that demanded leadership skills or political organizing. In the first post-war elections in Nepal in 2008, approximately a quarter of elected members of parliament (MPs) were formerly high-level party officials during the 1 For example, former rebel groups like SWAPO in Namibia, FSLN in Nicaragua, the NRA in Uganda, CNDD-FDD in Burundi, Fretilin in East Timor, Hamas in Palestine, among others, have held a majority of parliamentary seats in at least one post-conflict election. 71 war (Ishiyama and Marshall 2015). In El Salvador, the FMLN chose candidates “on the basis of historic command positions in the guerrilla organization” (Manning 2007, 265). The perceived relationship of the individual to violence also matters for how electable former rebels are. A key challenge for former rebel parties is convincing the public that they will adhere to institutional methods of political change and abandon violence (de Zeeuw 2008a). In general, rebel parties that directly declare an abandonment of the use of violence are more successful in gaining seats in parliament and in joining the executive (Ishiyama 2019). Moreover, rebel parties have also demonstrated concern with how the electorate experienced violence by the group during conflict (Söderström 2013). Ishiyama and Marshall (2015) demonstrate that in Nepal, the CPN(M) generally only ran former rebel members in their wartime strongholds, where the party had established positive relationships with the public and was unlikely to have been the target of CPN(M) violence. Thus, individual former rebels with heavy stigmas of violence may be unappealing political candidates. As rebel parties consider the electability of former rebels, I argue their female members theoretically possess advantages and disadvantages. In many groups, female rebels have the opportunity to gain skills and experiences that may qualify them for political office, particularly when they serve as combatants or leaders. On the other hand, there is an added negative stigma attached to violence by female rebels, as compared to males, that may hurt their electability. However, the role of gender has yet to be considered. As the state of the literature stands, it is unknown how common the political ascension of former female rebels is and how former rebel parties can reframe women’s participation and contributions. In the following section, I discuss the post-conflict lives of former rebel women. 3.3 The Political Viability of Female Rebels Interpretations of women’s participation in rebel groups in local communities fre- quently renders these women as lacking agency and political will, which could hurt their perceived political viability. Sjoberg and Gentry (2007) argue that rebel women are cast 72 as either “mothers,” “monsters,” or “whores.” With each category, the degree of social acceptability varies. Women are cast as “mothers” when their participation is seen as a sacrifice taken to liberate or fight for freedom in order to guarantee their family’s security (Sjoberg 2010). Their participation is out of necessity and closely tied to their role as moth- ers and protectors. In contrast, the “monster” narrative emphasizes that the women who join these movements and choose violence are psychologically disturbed, as the perpetration of violence defies social expectations for the role of women in society (Sjoberg and Gentry 2007). Women are assumed and expected to fall into the role of victim or peacemaker during wartime. When women deviate from these strict social norms, they are thought to be unnat- ural and immoral. Lastly, the “whore” narrative assumes that women in rebel groups have not only defied cultural expectations regarding sexual purity, but that they are hyper-sexual and thus sexually dysfunctional as a woman (Sjoberg and Gentry 2007). In many cases, women in rebel organizations are assumed to have served as camp wives or sexual slaves, regardless of their actual roles and responsibilities (Coulter 2008; Denov 2008). In southern Sudan, Deng Deng (2012) explains that in most communities, “the term WAAF [Women Associated with Armed Forces] is congruent with ‘loose women’” (167). These interpretations of women’s motives have consequences for how female rebels are perceived after conflict and subsequently determine their post-conflict marginalization. As a result, this framing affects their political viability, as well. Among women who are viewed as “monsters” or “whores,” they are often not only unable to build on their wartime experiences, but they often experience harsher discrimination and marginalization. When former female rebels return to their communities after war, they often become social pariahs and are frequently rejected by the broader community and even their friends and family (Coulter 2008; Denov 2008). For example, Black Diamond, a well-known former female commander of LURD in Liberia, has discussed the difficulty in finding acceptance after war, limiting her ability to find work or develop relationships, saying “they come up with lies about me, they make me look like a monster” (Nilsson and Thapar-Björkert 2013, 114). 73 In addition to the extreme social degradation, many former female combatants, like Black Diamond, also struggle to find sustainable work. In many contexts, female rebels have limited education or work experience to build on (Bop 2001).2 . These perceptions of former rebel women, and its real effects on their quality of life, likewise limit any political opportunities. For the rebel women who are cast as “mothers,” their participation is viewed as a nec- essary exception. Their participation is less likely to be sanctioned when they return home, though it will still be unlikely to receive the same veneration as men’s (Barth 2002). These categories—of “mothers,” “monsters,” and “whores”—highlight how the social acceptability of women’s violence is determined by the degree to which women adhere to gender norms. Sjoberg and Gentry (2007) argue that women are allowed to be agents when they act in a way that conforms to gendered expectations, such as in instances in which their political action is linked to motherhood or peace. For example, women fighting as soldiers within IS are seen as monsters or whores—misguided women who were brainwashed or kidnapped into the movement (Sjoberg 2018). In contrast, the women fighting in anti-IS forces are characterized as heroic and “fighting to protect their families,” (Sjoberg 2018). However, even when women’s participation in conflict is positively framed in terms of motherhood, it does not guarantee the ability to build on their wartime contributions in the political realm. Even women who are cast as “mothers” often find that when they return home, they are expected by both their communities and their former comrades to revert to traditional gender roles (Barth 2002). In Eritrea, Hale (2001) notes that throughout the war, female combatants experienced a sense of equality, yet “the rest of Eritreans held on to their identities and valued traditional familial and gendered roles” (138). Similarly, in Nepal, female combatants—many of whom were motivated to join the rebellion to fight 2 In sex-segregated labor markets, their ability to transform wartimes skills may be more limited than men. Former male combatants tend to find work more quickly than the women in skilled or semi-skilled work (Meintjes, Pillay, and Turshen 2002, 69). Formal rehabilitation programs often do little to mitigate these issues, frequently offering skill-building training that is gendered and lacks the opportunity for social or economic mobility (MacKenzie 2009) 74 for equality—found themselves returning to “the feudal, gender differentiated society that they had abandoned long ago” (Dahal 2015, 192). In consequence, even positive perceptions of women’s participation does not guarantee political opportunities. While perceptions of former rebel women may be less negative in some contexts, the individual level roles of these women had do not seem to guarantee opportunities after conflict. Instead, it may not matter just how these women are framed, but also who is framing them. For this reason, it is pertinent to look to the group-level to understand opportunities for former rebel women. Political parties generally serve as gatekeepers to women’s entrance into politics, fram- ing the political viability of female candidates and creating opportunities (Norris and Loven- duski 1993). Thus, we should expect that former female rebels’ ability to access political opportunity after war is likewise mediated by these parties. Specifically, former rebel parties have the opportunity to frame how the participation of their former rebel women is inter- preted, as well as give them the opportunities to overcome any marginalization. In the next section, I consider party-level explanations of the election of former rebel women. 3.4 The Election of Former Female Rebels Through their participation in rebel groups, I expect that many former female rebels could go on to establish post-conflict political careers. However, the opportunity to do, I theorize, is predicated on group-level factors, specifically women’s contributions to the orga- nization and how the rebel party perceives and frames these contributions. Narratives from former female rebels suggest that there is a consistent supply of women who are interested in politics after conflict. Former female rebels, like other rebels, will be unlikely to simply abandon any unmet political goals. Instead, through their former rebel parties, they have the opportunity and desire to continue fighting for these political ideals (Yadav 2016). They “move forward in a new formation,” working towards their political goals, simply using new tactics (Gowrinathan 2021, 124). Thandi Modise of the uMkhonto we Sizwe (MK) in South Africa famously told her male comrades, “We were next to you in the bush; we’ll be next 75 to you in government” (Turshen 2002, 86). Similarly, Hisila Yami of the CPN(M) argued that her choice to pursue a seat in parliament after war was a continuation of her political struggle, emphasizing that both revolution and her peaceful engagement in the Constituent Assembly are “features of political life” (NDI 2010, 17). While former rebel women consistently indicate an interest in politics, I expect that the post-conflict political fate of former rebel women may not necessarily be the result of their individual choices and interests after war, but instead is determined at the party-level. In the next section, I theorize how women’s contributions during war alter party perspectives about integrating women, ultimately making the group more interested in including women after conflict. The party’s former rebel women then offer a supply of women to integrate. I argue that in groups where women served as combatants and leaders, there will be a greater likelihood of improving the party’s attitudes on women’s inclusion and subsequently electing a greater percentage of former female rebels. Likewise, I expect that when women were included in the group and are seen as contributing specifically to victory, there will be a greater chance of altering attitudes of the party and thus, electing a greater percentage of former female rebels. However, I also expect that these altered attitudes could simultaneously hurt opportunities for former rebel women. As parties become more open and willing to elect women, they may be more likely to attract more qualified female candidates. In this case, I expect that rebels will opt to integrate other non-rebel women over their former members, and as such a lower percentage of former rebel women will hold party seats. 3.4.1 The Effects of Female Rebels’ Wartime Contributions As Turshen (2002) argues, for many rebels, their expertise lies in fighting, not politics. Scholars emphasize that rebel parties fill party seats with former members who held elite positions during war (Manning 2007). When women hold such elite positions in rebel groups during conflict, I argue that it is more likely that the party will view women as generally capable and qualified of continuing to hold elite positions in the party after war. Thus, groups that had women in more prestigious positions will be more likely to elect former 76 female rebels. Moreover, the battlefield achievements of these women may serve to improve perceptions of former female rebels among the public, which may also improve the overall political viability of former rebel women. I look to two primary roles that women fill within rebel groups—combatant and leader. At first glance, the role of combatant may be primarily related to training in related to combat, fighting, and weaponry. Yet, the role of combatant for women specifically is transformative for its perceived prestige. The role of combatant is be particularly empowering for how female rebels are viewed among the organization. Scholars have stressed how women’s training for these roles empha- sizes sameness. Men and women are trained in the same manner, and women find themselves succeeding in tasks they had traditionally been taught were reserved for men (Lahai 2015). This alters how these women view themselves, but also how their male comrades view them, ultimately garnering greater respect for these women soldiers (Barth 2002; Cock 1992). More- over, among the public, seeing women in similar roles as men may not only dissolve gender bias but may also challenge their stereotype of women in rebel groups as brainwashed or deviant to see them as valiant and brave. For example, Bop (2001) argues that women’s roles in war can transform views as women from “wives and mothers to that of fighters and liberators of their country” (21). Second, women’s roles as combatants also set them apart from other women generally. In particular, these roles are considered to be of high prestige (Thomas and Wood 2018). In general, most forces that include women assign them to gender-conforming support roles; the proportion of women fulfilling combat roles as compared to support roles is significantly smaller than it is for men (Thomas and Wood 2018). Thus, observing women in prestigious roles and seeing that they can excel in these positions may alter organizational views about who “elites” are, moving away from ideas that elite positions are meant only for males. Given that rebel parties generally turn to their most elite members to fill party seats (Manning 2007), as women perform in these leadership roles and alter perspectives of their group, this could influence who the party sees as fit for “elite” positions after conflict. Likewise, these 77 elite roles may also alter how former combatants, and specifically former female combatants are generally viewed by the public. As more women are seen in elite roles, the public may come to associate former rebel women with prestige. In Zimbabwe, Lyons (2004) argues that combat roles determined the political ascension of former rebel women. For example, Teurai Ropa Nhongo, quickly rose through the ranks of ZANLA and became one of the first female commanders in the rebel group. She became a skilled fighter and went on to lead the women’s detachment (Lyons 2004, 113). After the war, she was appointed as the Minister of Sports, Youth, and Recreation, becoming the youngest cabinet member ever. Further, she had no formal education among a cabinet filled with Ph.D.s; instead, she had a prestigious and “outstanding combat record” that qualified her (Seidman 1984, 430). Years later, she would be elected to parliament and ultimately serve as Vice President from 2004-2014. This leads to my first hypothesis: HYPOTHESIS 1: Rebel groups that included women in combat roles will be more likely to elect former female members after war. I also expect that groups that included women in leadership roles during conflict will be more likely to elect their former female rebels after conflict. Similar to how women’s roles as combatants could serve to alter group attitudes on women’s inclusion, as the group integrates women into leadership roles and observes their performance, they may come to see women as more fit for other leadership roles in the party after conflict. During conflict, women serve in a variety of leadership roles and perform important tasks, such as aiding to craft a public political identity, holding diplomacy positions, and serving as brokers to communities. These roles build important skills for women in the organization, as well as offer them the opportunity to build networks across the group—expanding the degree to which and to whom women demonstrate fitness for future roles. Leaders of rebel groups hold a variety of important responsibilities. They build intan- 78 gible things, like social endowments, within the group that include trust, shared values and beliefs, and overarching goals (Weinstein 2006). Further, leadership roles entail making hard decisions, delegating tasks, and navigating complex relationships with other actors. Certain wartime positions even demand political organizing that closely resembles the political orga- nizing parties take on after conflict. For example, Amrita Thapa was one of three top female leaders in the CPN(M) in Nepal during the people’s liberation war (IDEA 2011). During conflict, she served as the General Secretary of the Nepal Women’s Association and spent her time expanding the organization to all 75 districts of the country. She offered training in these districts on women’s rights and issues, relating it to the Maoist struggle (IDEA 2011). After her 2008 election to the Constituet Assembly, she claimed that it was her wartime political experience that prepared her to hold political office (IDEA 2011, 386). Yet, it was likely the example set by Thapa, who was just one of few women leaders during the war, that had a lasting affect on how the group integrated women after war. Despite only having a few women leaders, the CPN(M) filled more than 75 of their party seats with former rebel women in 2008, demonstrating that the altered attitudes may have been more influential than the individual skills built. Women’s relation to other high ranking rebel officials likely could also enable post- conflict opportunities. When considering women to back for political office, former rebel party leaders, like other political actors, will likely think to the qualified individuals within their networks. The more these leaders associate rebel women with leadership, the more likely they may be to incorporate other rebel women into the party after conflict. Following the Bush War in Uganda, former rebel leader—Yoweri Museveni—appointed a National Resistance Council that acted as the legislative body for the country. The only women appointed in its first year were former rebels (Mugabme 2000). A few of them served as leaders, but many served in support roles, also lending evidence that perhaps that altered attitudes through women’s general incorporation into leadership roles is more influential than the individuals’ background. This leads to my second hypothesis: 79 HYPOTHESIS 2: Rebel groups that included women in lead- ership roles will be more likely to elect former female members after war. While I argue that the roles of former female rebels within rebel parties will influence their election, I also expect that how these roles relate to the group’s success could determine the election of these women. In particular, this framing may lead the group to see women’s contributions as invaluable, and may also lessen the general bias held against rebel women by the public. First, in many cases in which former rebel groups were victorious, narratives emphasize the crucial role that women played in obtaining such victories. For example, in Uganda, Mama (2020) notes the ways in which women’s work as couriers, spies, and even cooks contributed to the NRA’s victory. As women’s roles in war are framed as necessary and consequential, groups may come to place greater value on the labor of the women in their organization and ultimately consider integrating them in greater levels after war. Second, I expect that rebel parties will elect a higher percentage of female rebels in contexts in which they victorious in the conflict because it will likely lessen the bias against these women. After war, the victorious have the opportunity to frame the conflict and their success, often reshaping legacies of violence and rewriting history (Daly 2019, 749). Lyons (2016) argues that these narratives tend to “recall the small, highly dedicated groups that initiated the rebellion, betrayal by traitors and rival factions, and the inevitability of their just cause” (170). How these narratives are framed similarly influences how former rebels are perceived in society. In cases of victory, former rebels can more easily frame themselves as liberators and freedom fighters, suggesting the use of force during conflict was a moral and just cause (Manning 2004, 2007).3 This framing of the conflict and former 3 In comparison, in cases of ceasefires, peace agreements, or even loss (though such occurrences are rare among rebel parties), it likely that rebel groups will be perceived differently. At the least, they will not necessarily have the opportunity to shape the narrative as much as rebels often do in cases of victory. Moreover, in these cases, it is less likely that rebels may tie the role of women to the conflict outcome, leaving little interactive effect between women’s contributions and the group’s settlement or even loss. 80 female rebels has relevant implications for how former female rebels are perceived. Moreover, in these moments rebels have greater opportunity to relate women’s participation in these movements to the motherhood narrative discussed by Sjoberg and Gentry (2007). In cases in which their respective rebel group was victorious in conflict and female rebels are framed as fighting to protect their country, it is the case that there is less discrimination and bias held against former female rebels, as their cause is seen as more just and moral. In contrast, the discrimination and marginalization of women who participated in fighting forces are more severe for those who are associated with groups that were defeated or are still perceived as illegitimate (Farr 2002). As such, female members of victorious rebel groups, like other former rebel fighters, are likely to be viewed as liberators and heroes, rather than as deviant or immoral. They become symbols of liberation and the liberated woman (Hale 2002, 124). For example, In Zimbabwe, women’s participation in the liberation forces was described as crucial to the struggle: Zimbabwean women have been playing a major role too in the fight for their country’s freedom . . . . They have shown in many difficult situations that their determination to liberate their motherland is second to none. As more of them get political education more will get more deeply committed than ever before. Zimbabwe needs everybody’s service. Zimbabwe shall and must be free. Our girls are playing a big part in this patriotic struggle.4 This framing has the potential to lessen the bias and discrimination that former female rebels face in the post-conflict environment. It also serves to depict former female rebels as just liberators fighting for their families and their country—ultimately making them more electorally viable, increasing the likelihood that their former rebel parties run these women and that the electorate opts to vote for them. This leads to my third hypothesis: 4 Thokozile Ushe. Zimbabwe Review. 1977 81 HYPOTHESIS 3: Rebel groups that won the conflict will be more likely to elect former female members after war. While I expect that women’s roles in rebel groups may influence greater levels of political integration of rebel women after conflict, it is possible that the changing attitudes facilitated through their engagement could actually also serve to hurt their opportunities in former rebel parties. I argue that as rebels observe the contributions of women members, they may be more open to electing women after conflict. This likely could result in rebel parties being more willing to recruit not only former rebel women, but women in general. This creates greater competition for former rebel women. The baseline bias held against women who commit violence may render them as non-preferable candidates compared to other non-violent women. This comparison may evoke narratives of Sjoberg and Gentry (2007)’s “monster,” for their use of violence. For rebel parties, former female rebels may be a risky investment. Moreover, for rebel parties that seek to distance themselves from a violent past, the public perception of former female rebels as ultra-violent or as “monsters” may serve to instead reinforce those associations to their history of conflict. In contrast, non- rebel women may offer rebel parties an opportunity to reframe their organization. Women generally are assumed to have greater inclinations to peace than men, particularly in post- conflict contexts (Berry 2018a; Tripp 2015). Thus, by recruiting more women, rebel parties have the opportunity to capitalize on these stereotypes of any women representatives and to demonstrate that they are, like their female MPs, dedicated to peace. Electing former female rebels is unlikely to have the same effect, as these women will be more likely to have to prove their dedication to peace, despite violent pasts.5 This could make them unviable candidates for former rebel parties—pushing parties to hesitate to support them. Instead, electing former female rebels may be a secondary preference to 5 However, this task may be more easy for some female rebels than others. For example, rebel women who participate in peace processes may have already had the opportunity to demonstrate that they are committed to upholding peace. 82 electing other non-rebel women. After conflict, there is a general push to increase women’s representation, from both domestic women’s movements and international actors (Tripp 2015). For this reason, rebels will be unlikely to simply replace former rebel women with former rebel men, and would instead turn to other women to fill certain seats. As such, when the party has a large supply of potentially qualified non-rebel female candidates, they may opt to run these women over former rebel women. Thus, while women’s participation in rebel groups may still alter perceptions at the party-level that leads rebel parties to be more willing and interested in electing women, rebel women may not be the preferable option to these parties in instances in which they are selecting which women to include. This leads to my fourth hypothesis: HYPOTHESIS 4: Rebel groups that have a smaller supply of qualified non-rebel female candidates will be more likely to elect former female members after war. 3.5 Research Design This paper argues that while former female rebels generally experience discrimination and marginalization after conflict (Burman and McKay 2007; Coulter 2008; Denov 2008; Doerrer et al. 2011; MacKenzie 2009; McKay 2005; Utas 2003), former rebel parties can mediate this bias and create political opportunities for former rebel women. I argue that the roles that rebel women held during conflict, including combat and leadership roles, can alter general attitudes about women’s inclusion, creating more “women-friendly” parties and leading to greater inclusion of former rebel women. I argue that this is particularly the case when former rebel women are seen as contributing the group’s victory. However, I also expect that former rebel women are the non-preferable female representatives for rebel parties, and that when there is a greater supply of qualified non-rebel women, a lower percentage of rebel women will be elected. To test these arguments, I introduce the Election of Former Female 83 Rebels Dataset, which documents the number of former female rebels elected to rebel groups that transformed into political parties. The sample includes all former rebel parties that participated in any post-conflict election between 1970 and 2020 and gained at least one seat in parliament. To identify former rebel parties, I relied on Daly (2020)’s Civil War Successor Dataset. The unit of analysis is party-year. Given the restriction of the sample to rebel parties, the analysis only considers the factors that lead to the inclusion of women in these parties and does not consider former female rebels running in non-rebel parties. The sample is also restricted to only include former rebel parties that participated in conflicts in which included females were included in a rebel group, as in order to elect former female rebels, there must be a supply of female rebels in the given context. To identify rebel groups that included female members, I first identified all rebel groups that engaged in civil conflict using UCDP’s Armed Conflict Dataset. I then relied on data from Thomas and Bond (2015), which records women’s membership in African rebel groups6 , and Henshaw (2016), which identifies female membership in a random sample of rebel groups. For the remaining groups that are not included in either of the respective samples, I coded female membership following the definitions of female membership outlined by Thomas and Bond (2015) and Henshaw (2016), which identify female members as those that “offered willful engagement in the activities, administration, or maintenance of the organization,” (Thomas and Bond 2015, 492) and that “clearly identify with the goals, ideology and/or efforts of one side in a conflict” and either serve in support roles in which they offer “labor, supplies, and assistance” or serve in combat roles in which they “regularly serve in frontline operations” (Henshaw 2016, 44). After identifying the sample of rebel parties with female members, I coded individual data for all female representatives elected in a post-conflict election that rebel parties in the sample won at least one seat in. I coded background information, including rebel membership, for a total of 1,956 women. By coding individual data on each female representative, I was able to 6 This differs from the Women in Armed Conflict Dataset (WARD) from Wood and Thomas (2017), which includes combatants, rather than members. 84 determine if each woman had wartime affiliations or not. This was necessary, as a significant portion of women elected to former rebel parties are not former rebels. Of the women in the sample, approximately 30 percent are former rebel women.7 0.30 0.25 0.20 P roport ion of Observat ions 0.15 0.10 0.05 0.00 0.0 0.1 0.2 0.3 0.4 0.5 0.6 P rop. Women's Seat s (By Part y) Figure 3.1: Proportion of Women’s Seats The dataset includes a total of 51 unique rebel parties participating in 200 elections throughout 34 countries. Rebel parties in the sample participated in on average 4 elections, with some parties even participating in as many as 9 elections across the temporal scope. On average, rebel parties in the sample won approximately 34 percent of parliamentary seats in each election. Of the 51 rebel parties in the sample, 20 have at some point held a majority of seats in parliament. On average, each party filled about 20 percent of their seats with women.8 Figure 3.1 shows the average proportion of women’s seats at the party level across the sample. Figure 3.2 shows the average proportion of party representatives elected who are former female rebels. In approximately half the sample, no female rebels were elected. However, in a number of cases, former female rebels hold a significant number of seats, 7 For a more detailed discussion of the coding of these data, please see Appendix C. 8 This is higher than the national average of women elected across all parties, which is about 18 percent. 85 0.60 0.70 0.55 0.65 0.50 0.60 0.45 0.55 P roport ion of Observat ions P roport ion of Observat ions 0.50 0.40 0.45 0.35 0.40 0.30 0.35 0.25 0.30 0.25 0.20 0.20 0.15 0.15 0.10 0.10 0.05 0.05 0.00 0.00 0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 P roport ion of Part y Seat s Held by Female Fight ers P roport ion of Women's Seat s Held by Female Fight ers Figure 3.2: Prop. Former Female Rebels Figure 3.3: Prop. Former Female Rebels (Party) (Women’s Seats) making up more than 10 percent of all representatives in about 15 percent of observations. Figure 3.3 shows the average proportion of women elected who are former female rebels. In about 50 percent of elections, none of the women elected to the rebel party are former female rebels. Yet, in cases in which former female rebels are elected, they make up a significant proportion of the women representatives. In about 25 percent of elections, former female rebels represent at least 25 percent of the women elected. In 15 percent of elections, more than 50 percent of the women elected are former female rebels. Finally, there are a handful of cases in which rebel parties only elect women who are former female rebels. For example, in the first election following the peace agreement in El Salvador, the FMLN only elected women who were former rebels. The dataset also presents interesting regional variation in the inclusion of former female rebels. On average, parties in Latin America elect the higher percentage of former rebel women, with former rebel women representing approximately 53 percent of all women elected and 17 percent of the party. Following Latin America, the second and third highest percentage of former rebel women in parliament are found in Africa and the Middle East, with 86 each having rebel women represent about 14 percent of their women, but just 3 percent of the total party. On average in Asia, former rebel women represent about 8 percent of women elected and just 3 percent of the party. Finally, Europe has the lowest average percentage of former rebel women elected, with these women holding just 3 percent of women’s seats and 1 percent of party seats. The dataset itself is a significant contribution to the literature on gender and conflict, as to my knowledge, there have previously not been any efforts to collect large-N data on the election of former female rebels. Thus, this dataset serves to establish several basic understandings of when, where, and why former female rebels are elected after conflict. This is of relevant consequence for understanding the exclusion of female rebels as compared to male rebels and subsequently for understanding which women are gaining opportunities after conflict and which women are further marginalized and could stand a threat to peace stability. Moreover, the dataset can well complement rich, qualitative work that has been done on the political lives of former female rebels to demonstrate the political marginalization or integration of these women. 3.5.1 Dependent Variable The dependent variable is the proportion of former female rebels elected in a given rebel party in an election year. This variable is calculated by dividing the number of female rebels elected to the party by the total number of women elected to the party in a given election. I opt to specify the variable this way, as opposed to operationalizing it as the total number of female rebels divided by the total number of party seats because I assume that the number of female rebels that fill seats will be dependent on how many females they elect overall. This assumption rests on the idea that the party approaches the recruitment of female candidates differently than they do male candidates, which scholars have found evidence of in other contexts (Krook 2010; Kunovich and Paxton 2005; Niven 1998; Sanbon- matsu 2002). From this perspective, women’s access to politics is dictated by the willingness of the party (and the male gatekeepers in the party) to accept women as legitimate political 87 actors (Valdini 2019). Thus, the degree to which former rebel women are integrated into rebel parties will depend on the party’s overall willingness to accept women. In contrast, it is possible that former rebel women’s inclusion to rebel parties is more dependent on the party’s willingness to run former members. I use an alternate specification of the dependent variable to account for this possibility in Appendix C. 3.5.2 Explanatory Variables The first key explanatory variable is Female Combatants. This is used to test Hy- pothesis 1, which suggests that rebel parties that had female combatants will elect a higher percentage of former female rebels. I operationalize this in two ways. The first is through a dichotomous variable that indicates whether or not a given rebel party had female combat- ants during conflict. The second is through an ordinal variable that indicates the proportion of combatants in a given rebel group that are female. These variables come from the Women in Armed Conflict Dataset (WARD) (Wood and Thomas 2017). I opt to use two opera- tionalizations in order to understand if it matters simply that women were combatants, or if it matters the extent which women were integrated as combatants. The second explanatory variable, Female Leadership, is used to test Hypothesis 2, which argued that rebel parties that had female leaders will elect a higher percentage of for- mer female rebels. This is a dichotomous variable from Henshaw et al. (2019) that indicates whether or not the group had female leaders during war. My third explanatory variable, Rebel Victory, is used to test Hypothesis 3, which argues that when rebel parties are victori- ous in conflict, they will elect more former female rebels. This is a dichotomous variable from the Non-State Actors Dataset (Cunningham, Gleditsch, and Salehyan 2009) that indicates whether or not a group won its conflict. Finally, to test Hypothesis 4, which argues that rebel parties operating in contexts with a greater population of qualified women will elect fewer former female rebels, I rely on two operationalizations. First, I use a variable on women’s workforce participation, which comes from the Varieties of Democracy that estimates the access of women to state jobs. The variable is an ordinal measure that rates women’s access 88 to state jobs as either unequal, somewhat unequal, relatively unequal, or equal. I opt to use this operationalization over other variables that estimate women’s workforce participation as women holding state jobs would have more relevant qualifications for holding political office than the average women. As women hold higher positions within these fields, more may gain professional networks and experiences that well position them to run for office. Second, I use women’s civil society participation to proxy for qualified female candidates, as many women enter politics via civil society engagement and leadership and these organizations sometimes even offer candidate training programs (Yoon 2004). To measure this, I use a variable from the Varieties of Democracy dataset that measures the level of women’s exclusion from civil society organizations, measuring their exclusion as frequent, about half the time, rare, or almost never (Coppedge 2019). I include several relevant control variables. I include a variable for GDP per Capita (logged) from the Maddison Project, as the literature has demonstrated that countries with higher GDP per capita generally have higher levels of women’s representation (Hughes and Paxton 2008; Paxton, Kunovich, and Hughes 2007c). I include a variable for the level of democracy, as the literature has generally demonstrated that democracies are more likely to elect women (Bush 2011; Phillips 1995). Likewise, scholars have also demonstrated the relationship between autocracy and women’s political representation (Donno and Kreft 2019; Valdini 2019). I control for the adoption of gender quotas, as this is associated with higher levels of women’s political representation (Hughes and Paxton 2008; Paxton, Kunovich, and Hughes 2007b). Finally, I control for women’s fertility levels, in order to capture a sense of gender roles in the country, as in places that expect women to hold traditional roles, such as mothers, may be more disinclined to favor women transgressing these norms by joining rebel groups. To measure fertility, I use data from the World Bank which estimates the average number of children had by women in a given country. 89 3.5.3 The Model To understand the determinants of the election of former female rebels, I utilize a tobit maximum likelihood estimator that is left-censored at zero. Given the significant number of observations (nearly 50 percent) in which former rebel parties elected zero former female rebels, a tobit model is appropriate, as these models account for bias that arises when a large proportion of the data is clustered at one end of the distribution (Wooldridge 2010, 2012). Further, a tobit model is useful in this analysis given that the analysis does not utilize candidacy data and only considers if former female rebels were elected. It can be assumed that some rebel parties run no former female rebels and thus elect no former female rebels, while other parties may run former female rebels, but actually elect no former female rebels. Running a linear model that assumes that these two processes are the same would lead to bias in the model. In contrast, tobit models account for such bias by accounting for the fact that there is a large and disproportionate amount of data clustered at one end of the observable data (Wooldridge 2012). The unit of analysis in the model is party-year. Standard errors are clustered by country in all models. 3.6 Results Table 3.1 presents the results of the tobit analyses. In Models 1 and 2, the coeffi- cients for female combatants and the proportion of female combatants are each positive and significant. This lends support to Hypothesis 1, which suggests that rebel parties that had female combatants will fill a higher percentage of their seats with former female combatants. Figure 3.4 shows the marginal effect of including female combatants, estimating that rebel parties that had no female combatants will fill nearly zero percent of their seats with former female rebels, while those who had female combatants during conflict will fill nearly 20 per- cent of their female seats with former female rebels. Figure 3.5 shows the marginal effect of the proportion of female combatants that rebel parties had during conflict. I estimate that in groups where women represented only 5 percent or less of fighting forces, former female 90 Table 3.1: Tobit Model Dependent Variable: % Former Female Rebels Elected (1) (2) (3) (4) (5) (6) (7) Female Combatants 0.810∗ 0.793 0.843∗ 0.591 (0.405) (0.421) (0.380) (0.399) Prop. Female 0.427∗∗∗ 0.383∗∗∗ Combatants (0.077) (0.080) Female Leaders 0.463 0.371 0.062 (0.329) (0.216) (0.228) Women’s Access to -0.338∗ -0.231 0.0137 State Jobs (0.158) (0.133) (0.095) Women’s Civil -0.384∗ -0.349∗∗ -0.305∗ Society Participation (0.177) (0.131) (0.122) Level of Democracy 0.448 0.296 0.591 0.814 0.0637 0.718 0.156 (0.660) (0.471) (0.668) (0.529) (0.653) (0.540) (0.390) log(GDP per Capita) 0.141 0.245 0.041 0.155 0.271 0.198 0.297 (0.183) (0.148) (0.150) (0.191) (0.172) (0.165) (0.154) Gender Quotas -0.098 0.105 0.001 -0.034 -0.067 0.030 0.126 (0.228) (0.179) (0.241) (0.183) (0.170) (0.169) (0.136) Rebel Victory 0.264 -0.031 0.058 0.413 0.163 0.246 -0.122 (0.335) (0.214) (0.326) (0.271) (0.264) (0.203) (0.191) Fertility Levels 0.094 0.150 0.155 0.0646 0.131 0.113 0.175 (0.118) (0.091) (0.101) (0.126) (0.110) (0.125) (0.097) Constant -2.673 -3.529∗ -1.631 -2.609 -3.066 -2.670 -3.440∗ (1.902) (1.617) (1.530) (2.030) (1.827) (1.837) (1.638) Sigma 0.312∗∗ 0.167∗∗ 0.287∗∗∗ 0.243∗ 0.250∗ 0.197∗ 0.132∗ (0.107) (0.061) (0.081) (0.095) (0.100) (0.089) (0.054) No. Obs. 111 110 120 111 111 110 110 Pseudo R2 0.133 0.393 0.105 0.237 0.203 0.306 0.459 Log Likelihood -65.34 -45.53 -76.59 -57.56 -60.12 -52.00 -40.53 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 rebels will fill less than 5 percent of the seats held by women with former female rebels. This could be both because the supply of female rebels is low, as well as because in these cases women may have been unlikely to have been central to accomplishing the aims and goals of groups. However, it is not possible to distinguish between the two explanations. In contrast, in groups with high levels of female combatants, in which women made up more than 20 percent of fighting forces, I estimate that of the women elected in these parties, 50 percent will be former female rebels. 91 Figure 3.4: Marginal Effect of Female Com- Figure 3.5: Marginal Effect of the Propor- batants tion of Female Combatants In Model 3, the coefficient for female leadership during conflict is positive, but in- significant. This does not offer support for Hypothesis 2, which suggests that rebel parties that had female leaders during conflict will elect a higher percentage of former female rebels. Hypothesis 3 suggests that rebel parties that were victorious in conflict will be more likely to elect their former female rebels. However, Models 1-6 offer no support for this theory. The direction of the coefficient is inconsistent across models and is insignificant in each. Hypothesis 4 posited that rebel parties that operate in contexts in which there is a greater population of potentially qualified female candidates will elect fewer former female rebels. The models offer mixed support. Women’s access to state jobs has a negative and signif- icant effect in Model 4, but is insignificant in Models 5-7. Figure 3.6 plots the marginal effect of women’s access to state jobs on the proportion of former female fighters who are elected. When women’s access is estimated at its lowest point (unequal), I estimate that approximately 38 percent of women elected will be former rebel women. In comparison, I estimate that when women’s access to state jobs is moderately high (relatively unequal), only about 10 percent of women elected will be women. Women’s civil society participation 92 Figure 3.6: Marginal Effect of Women’s Access to State Jobs has a negative and significant effect in Models 5-7, suggesting that when women’s civil so- ciety participation is higher, fewer rebel women are elected. Given the heavy involvement of women’s civil society groups in encouraging women to run for office, and often offering training programs and resources for candidates (Wang and Muriaas 2019), this finding is aligned with the theory when more women are qualified for political office, former female rebels will be less likely to successfully run for office. Figure 3.7 plots the marginal effect of women’s civil society participation. The plot shows that when limitations to women’s civil society participation are higher (about half the time), the relationship to the proportion of former rebel women is insignificant. However, I estimate that when women are excluded from civil society organizations about half the time, about 65 percent of the women elected to the party will be former rebel women. In contrast, I estimate that when women’s civil 93 Figure 3.7: Marginal Effect of Women’s Civil Society Participation society participation is at its highest, meaning that women are almost always integrated into civil society, only about 20 percent of women elected to the party will be former rebel women. Among the control variables, there are few significant effects on the relationship spec- ified. The level of democracy has a positive but insignificant effect. GDP per capita has a consistently positive effect in all models, but is insignificant. The coefficient for gender quotas switches directions between models but is also insignificant in each model. The coefficients for fertility levels are positive, but insignificant in each model. 3.6.1 Conflict and Post-Conflict Factors I consider several relevant factors related to the conflict that may also influence the election of former female rebels. First, I consider whether groups used forced recruitment tactics during war, as the forced recruitment of women into the group likely imply a lack of autonomy in many of the actions and experiences of women in these organizations and 94 decrease their interest in continuing to work within these parties. This dichotomous variable indicates whether or not groups used forced recruitment tactics and comes from Henshaw et al. (2019). Model 1 in Appendix C shows that while forced recruitment tactics have a negative effect, it is insignificant. I also consider whether groups had a political wing, as it is conceivable that this would offer even more women the opportunity to gain relevant political experiences. To test this, I use a dichotomous variable for whether or not a given rebel group had a political wing from the Non-State Actor Dataset (Cunningham, Gleditsch, and Salehyan 2009). I find that the presence of political wings has a negative, but insignificant effect. Though surprising, it is possible that the presence of a political wing also increases the number of qualified male fighters, leading to more male rebels filling seats over women. I consider the effect of rebel governance during conflict, as it is possible that these governance structures offer rebel women greater opportunity to gain relevant political experience. I use two variables from the Rebel Governance Dataset (Huang 2016). The is a count variable that measures how many institutions rebels offered during conflict. The second is a dichotomous variable that indicates whether or not rebels had international diplomacy efforts during conflict. I find that each of these variables has a positive and significant effect on the percentage of rebel women elected. Next, I consider the role of ideology, as rebel groups with gender-positive ideologies may be more likely to integrate women after conflict. I use data from the FORGE Dataset that indicates whether groups had a nationalist, leftist, or Islamist ideology (Braithwaite and Cunningham 2020). Model 5 in Table C3 of the Appendix shows that nationalist and leftist ideologies have a positive, but insignificant effect. Islamist ideologies have a negative and significant effect. The primary results remain consistent with the inclusion of these variables. I also consider a few factors relevant to conflict resolution. I test for the presence of UN peacekeepers, as international influence could effect women’s general inclusion in politics. This variable comes from the UCDP Conflict Termination Dataset. Model 7 in Table C3 in Appendix C shows that the presence of UN peacekeepers has a positive, but insignificant 95 Figure 3.8: Marginal Effect of Time effect. Next, I test for the effect of electoral participation provisions, which legally allow for former rebels to run for elected positions. This dichotomous variable comes from Matanock (2016). The variable has a positive, but insignificant effect. I next consider if other conflict resolutions beyond rebel victories matter. Using data from the UCDP Conflict Termination Dataset, I test for the effect of peace agreements in Model 9 in Table C4. I find that in these instances, rebel parties elect significantly more female rebels, suggesting that these processes in some way lessen bias against female rebels. It is possible that as parties and the public see rebel women acting in peaceful, political roles through these processes, they come to further see these women as capable political actors. I also consider the effect of time, as the degree of integration of former female rebels may lessen with time as the party evolves and continues to recruit new members. Moreover, after several decades since conflict resolution, many of these women may have redirected their careers or are even retired. Figure 3.8 shows the marginal effect of the number of years since conflict. It estimates that the highest level 96 of former female rebels will be elected directly following conflict. With time, the percent of female rebels elected decreases substantially. This is logical, as over time the party may expand its networks to recruit more heavily beyond wartime members and as the former rebel population increases in age, they may retire or move onto other careers. Throughout all of these specifications, the primary results remain consistent. Finally, I consider aspects of party strength that could influence the election of women to rebel parties, as stronger parties with more resources tend to elect higher proportions of women. In Table C4 in Appendix C, I test for the effect of seat share and opposition party status. Seat share is measured as the proportion of seats each party holds in the legislature. These data on seat share were collected from the Inter-Parliamentary Union. In Model 11, the effect of seat share is positive and significant. The results remain consistent with the addition of this variable. In Model 12, I include a variable for opposition party status. This variable is based on data from the Varieties of Democracy V-Party Dataset (Lührmann and Al 2020). Accounting for opposition parties ensures that the relationship is not driven by dominant party status, nor that parties with more resources and access to state funds are more likely to use these strategies and to succeed in electing women. I find that opposition party status has a positive, but insignificant effect. The primary results remain consistent. 3.6.2 Robustness Checks I conduct several robustness checks. First, I use an alternate specification of the dependent variable. The main models operationalize the dependent variable as the number of former female rebels elected divided by the total number of women elected to the party. In Models 13-19 in Appendix C, I use an alternative operationalization, which is the proportion of former female rebels elected as calculated as the number of former female rebels elected divided by the total number of party seats. The results remain consistent across all models. I next use alternate operationalizations for a handful of control variables. Models 20-22 in Appendix C specify the model using data from Polity2 to measure democracy, as opposed to V-Dem. Models 23-25 use data from the World Bank for GDP per Capita, instead of 97 the Maddison Project. In Models 26-31 in Appendix C, I specify the model using alternate variables for gender quotas. Rather than using a variable that specifies if a gender quota has simply been adopted, I test for the actual implementation of gender quotas and whether the gender quotas were effective. The results remain consistent under these specifications. Finally, I use a few alternative model specifications. In Model 32, I use a linear regression model instead of a tobit model to estimate the results. In Model 33, I add fixed effects for time.9 The results remain consistent under these specifications. 3.7 Conclusion Scholars have found that the experiences of former rebel women after war are dom- inated by marginalization, discrimination, and negative stigmas. These women are harshly punished for their violent deviation from traditional gender norms and find their post-conflict opportunities limited. In this paper, I push back against the notion that this is a universal experience. I ask if some former rebel women, like their male comrades, can successfully turn their wartime experiences and connections into political capital in instances in which their former rebel groups transform into political parties. The data presented indicates that this is possible, as former rebel women are elected in significant numbers to many rebel parties. I find that their integration into these parties is determined by the roles of women during conflict, as well as by the overall status of women in the country. In cases in which women served as combatants during war, rebel parties are likely to elect a higher percentage of former female rebels. However, there is some evidence that in contexts in which there is potentially a greater supply of qualified and motivated population of women who could successfully run for office, fewer former female rebels are elected. This suggests that while former female rebels can overcome bias in some cases, they are still not the preferred women for these political positions. 9 I do not include country fixed effects, as there is often only one rebel party operating in a given country and many of the explanatory variables are time-invariant. 98 In demonstrating that not all former rebel women’s post-conflict lives are limited by the stigma of war and violence, my intention is not to negate the experiences of women’s whose experiences are defined by this marginalization. Instead, I intend to demonstrate how wide the gaps are among former rebel women. Many women find themselves worse off after participating in rebellion. This paper demonstrates that a select few will find themselves far better off. This echoes work by Giri (2021), Yadav (2017), K.C (2019), and K.C and Van Der Haar (2019) on the post-conflict lives of former female members of CPN(M) in Nepal, which demonstrates that the post-conflict fate of former female rebels is heavily dictated by intersectional identities. Those whose lives improve are often those who were already wealthy, educated, and well-connected. This demands that as we consider how to rectify the marginalization, we also aim to lessen the inequalities. This paper offers a broad, global understanding of where, when, and why former female rebels take on political careers after conflict. Such an understanding leads to several further questions that ought to be examined in future work. First, the paper does not consider candidacy processes. Moreover, given that former female rebels carry with them at least two identities that could influence their electoral success—their gender and former rebel status—their experiences as candidates may be unique. Future work ought to examine how wartime experiences are highlighted or perhaps minimized in campaigns and if public bias is a significant barrier to entering politics for these women. Second, given that former female rebels are in some cases elected in great numbers, how do they influence policy? Future work ought to examine the policy priorities of these women and how their wartime experiences influence the policies they pursue and support. 99 CONCLUSION 4.1 Summary of Findings In this dissertation, I set out to understand the role of women in rebel parties. I argue that the incorporation of women into rebel parties is strategic. Women serve as symbols for these parties, signaling that the party is more just, moral, and most pertinently, peaceful than they have been in the past. Their incorporation is used to boost support, and thus success. I argue that when and to what extent women are incorporated is influenced by the conflict legacies of these parties, including their need to distance themselves from a violent past, as well as a need to carry on the support of their wartime loyalists. I find that women’s political representation is unique in former rebel parties. Women are incorporated into these parties at significantly higher levels than they are in other political parties. There is some variation among rebel parties, however. Rebel groups that included women during conflict are even more likely to elect higher percentages of women to their parties after their transition. This is particularly true in instances in which women have shown before to offer a strategic benefit by increasing domestic and international levels of support. There is also significant variation in which women rebel parties elect. Consistent with their other practices, rebel parties are most likely to include their former rebel women when again it does not come at a cost. Former rebel women are unlikely to be the first choice as candidates for these parties—rather rebel parties include them when they can be framed in a way that sheds a positive light on the group’s wartime legacies. 100 4.2 Implications The findings in this dissertation demonstrate the persistence of conflict legacies, even decades following the conclusion of war. With time, women’s integration into rebel parties continues to grow. Moreover, the extent to which rebel parties regurgitate narratives of women’s contributions during conflict and tie them to the present contributions of women in politics is also appearing to grow. On International Women’s Day in 2020, Ugandan president and former rebel leader, Yoweri Museveni, tweeted, “Women are the base of so- ciety. Old society decided to marginalize and suppress them. This has changed with the recruitment of women in the armed forces since 1986 by the NRA. Happy Women’s Day.” The irony of Museveni’s tweet, along with other more recent statements praising women’s engagement in the Bush War, is that in the first decade or so following the war, he seldom acknowledged women’s participation, let alone the key role that they played (Tripp 2000). However, as women’s political inclusion has become more popular among the public, and among the international community, leaders like Museveni have taken notice. As such, they have successfully learned how to manipulate these narratives for their own gain. The endur- ing need to tie women to these conflict legacies suggests the ways in which these actors are still, even decades after war, trying to communicate that they are just and peaceful. The use of women in this framing offers an understanding of the shallowness of the “add women and stir” approach. Women are used strategically in these moments to signal that these parties are peaceful, but it does not seem to the case that this is actually true. In fact, it appears that many of the former rebel leaders who have relied on these strategies, like Museveni in Uganda or Kagame in Rwanda, are also trying to use them to continue distracting from ongoing violence and repression (Berry 2018a). This points to the absolute necessity that scholars, activists, and state leaders cease using the presence of women as a metric for peace or stability. Advocating for presence without the accompaniment of actual influence leads to manipulation of women’s inclusion for the benefit of often unjust leaders. 101 This leads to a related consideration of how peace and stability are considered in post-conflict countries. Relatedly, this body of work points to the importance of considering which women are receiving these post-conflict opportunities. If rebel parties can manipulate the election of women, as I have argued that they have, it is quite feasible that which women run for them is also manipulated. Recent work on women’s rights after war has demonstrated that it is often a select group of women who actually experience gains after war, not all women (Berry 2018a; Berry and Lake 2017, 2020). Interestingly, my findings on the election of former rebel women suggest that it may not be the women with the closest networks to rebels who are given opportunities after war. Instead, rebel women are elected in much smaller numbers than other women, which I argue is due to shortcomings in the electoral viability of these women. The introduction of new women to these rebel parties could offer a more positive trend, in which new voices are in fact entering the arena and could likewise push rebel parties’ policies in new directions. At the same time, without a greater understanding of which new women they are recruiting, it is hard to know what and who these women stand for, as the election of some women will not guarantee all women are represented. Finally, this dissertation offers a more nuanced understanding of the post-conflict fate of former rebel women. The findings demonstrate that select women do overcome marginal- ization and bias to take on political positions. This positively suggests that there is not a dramatic gender gap in the ability of former rebels to build on their wartime experiences. However, it does point to the fact that there are dramatic gaps among rebel women, as some women do in fact experience those harsh and reverberating social sanctions for their wartime participation, while others are able to take on some of the most prestigious political positions in their countries. These gaps are especially relevant for post-conflict peace building. Dis- armament, Demobilization, and Reintegration (DDR) programs have been heavily criticized for their failure to incorporate a gendered perspective into their programming. However, these findings suggest that in addition to gendered programming, international actors need 102 to learn how to address the gaps that emerge among former rebels. 4.3 Future Research While writing this dissertation, it became increasingly apparent that many projects can be derived from this work. I will briefly describe a few of the most pressing. First, the findings pointed to the fact that we must understand if and how women representatives influence rebel parties. It could be the case that women’s political representation is manip- ulated in these contexts and these women have little real influence over policy. However, it may also be possible that women could be creatively finding ways through which to grow their own influence in these spaces and to create change. The distinction between the two is important for how women’s political representation is viewed in these parties—essentially rendering it shallow or sincere. Therefore, more work must be done to understand women’s real policy preferences and work within these parties. Second, this dissertation is faulted by gaps in our understanding of women’s candidacy in rebel parties. Much of these gaps are the result of the difficulty in accessing data on women’s candidacy. As a result, future research should begin by at least focusing on one or several cases and offering an in-depth examination of women’s candidacy in rebel parties, including their recruitment by rebel parties, as well as any campaign support. Additionally, within this work is needed an understanding of how rebel parties and former rebel candidates specifically are viewed by the electorate. How is the association of the individual to violence perceived by the public and how is this gendered? More research must be done here to understand how candidate recruitment, campaigning, and voter bias may be gendered for former rebels. Finally, this dissertation highlighted the political engagement of female rebels after war. Much of the work on the post-conflict lives of female rebels are dominated by narratives of marginalization. The findings presented here offered more nuance to this conversation. However, this is only a start. More work must be done to understand which former rebel 103 women are getting these opportunities and why. Further, in writing that chapter, it became apparent that many former rebel women engage in other political activities after war, often after experiencing neglect or rejection from their former organization. In what other ways do former rebel women engage politically after war and how is this influenced by both their war and post-war experiences? Lastly, in uncovering the nuance in experiences among former rebel women, the inequality in opportunities was striking. A critical area, thus, to be developed is how these gaps emerge and more importantly, how they can be remedied. 104 APPENDICES 105 APPENDIX A APPENDIX TO CHAPTER 1 Table A1: Descriptive Statistics Statistic N Mean St. Dev. Min Pctl(25) Pctl(75) Max Percent Women Candidates 148 0.30 0.14 0 0.19 0.40 0.61 Percent of Women’s Seats 847 0.10 0.20 0 0 0.1 1 Rebel Party 847 0.14 0.35 0 0 0 1 Democracy 847 0.53 0.50 0 0 1 1 Gender Quotas 847 0.39 0.49 0 0 1 1 Exclusive Candidate Selection 847 0.62 0.49 0 0 1 1 Opposition Party 847 0.12 0.33 0 0 0 1 Party Institutionalization 844 0.48 0.17 0.15 0.34 0.58 0.85 log(GDP per Capita) 811 7.38 0.95 6.24 6.55 8.26 9.41 Women’s Civil Society Participation 847 0.56 0.18 0.20 0.44 0.72 0.89 106 Table A2: Multilevel Linear Regression Model (Time) % Women Candidates % Women Elected (1) (2) (3) (4) Rebel Party 0.07∗∗ 0.08∗∗∗ 0.06∗ 0.08∗∗∗ (0.02) (0.02) (0.03) (0.02) Level of Democracy −0.11 −0.01 −0.01 −0.00 (0.06) (0.02) (0.02) (0.02) Gender Quotas −0.08 0.04∗ 0.04∗ 0.04∗ (0.05) (0.02) (0.02) (0.02) Candidate Selection −0.01 −0.01 −0.01 −0.01 (0.05) (0.02) (0.02) (0.02) Opposition Party −0.02 0.02 0.02 0.02 (0.03) (0.02) (0.02) (0.02) Party Institutionalization 0.51∗∗ 0.16∗ 0.16∗ 0.14 (0.17) (0.07) (0.07) (0.07) log(GDP per Capita) −0.00 −0.02∗ −0.02 −0.02 (0.03) (0.01) (0.01) (0.01) Women’s Civil Society Participation 0.35∗ 0.16∗∗∗ 0.16∗∗∗ 0.17∗∗∗ (0.18) (0.05) (0.05) (0.05) Time −0.01∗∗∗ 0.00∗∗∗ 0.00∗∗ 0.00 (0.00) (0.00) (0.00) (0.00) Rebel Party∗Time 0.00 (0.00) Time2 0.00 (0.00) Constant −0.16 0.04 0.04 0.02 (0.19) (0.09) (0.09) (0.09) Country Variance 0.00 0.00 0.00 0.00 AIC −104.20 −411.99 −401.11 −395.17 Log Likelihood 67.10 220.99 216.55 213.59 No. Obs. 124 808 808 808 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Time splines omitted 107 Table A3: Multilevel Linear Regression Model (Candidacy & Conflict Outcome) % Women Candidates (5) (6) (7) Rebel Party 0.06 0.07∗ 0.07∗ (0.04) (0.03) (0.03) Democracy −0.06 −0.05 −0.01 (0.12) (0.14) (0.11) Gender Quotas −0.05 −0.05 −0.02 (0.06) (0.06) (0.07) Candidate Selection −0.03 −0.03 −0.04 (0.06) (0.06) (0.06) Opposition Party −0.02 −0.02 −0.02 (0.03) (0.03) (0.03) Party Institutionalization 0.12 0.10 0.09 (0.25) (0.25) (0.23) log(GDP per Capita) 0.03 0.04 0.06 (0.05) (0.05) (0.05) Women’s Civil Society Part. 0.13 0.12 −0.15 (0.32) (0.31) (0.42) Rebel Victory −0.02 (0.09) Rebel Party*Rebel Victory 0.01 (0.05) Peace Agreement −0.00 (0.15) Rebel Party*Peace Agreement −0.02 (0.05) Low Activity −0.17 (0.21) Rebel Party*Low Activity −0.03 (0.11) Constant −0.11 −0.14 −0.11 (0.34) (0.38) (0.33) Country Variance 0.01 0.01 0.01 AIC −97.53 −98.91 −101.81 Log Likelihood 64.76 65.45 66.90 No. Obs. 124 124 124 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Time splines omitted 108 Table A4: Multilevel Linear Regression Model (Electoral Success & Conflict Outcome) % Women Elected (8) (9) (10) Rebel Party 0.07∗∗ 0.07∗∗ 0.10∗∗∗ (0.02) (0.02) (0.02) Democracy −0.02 −0.02 −0.03 (0.02) (0.02) (0.01) ∗ ∗ Gender Quotas 0.04 0.04 0.04∗ (0.02) (0.02) (0.02) ∗ Candidate Selection −0.04 −0.04 −0.04 (0.02) (0.02) (0.02) Opposition Party 0.02 0.02 0.03 (0.02) (0.02) (0.02) Party Institutionalization 0.25∗∗∗ 0.24∗∗∗ 0.25∗∗∗ (0.07) (0.07) (0.07) ∗ ∗ log(GDP per Capita) −0.03 −0.03 −0.03∗ (0.01) (0.01) (0.01) ∗∗∗ ∗∗∗ Women’s Civil Society Part. 0.18 0.21 0.19∗∗ (0.05) (0.05) (0.06) Rebel Victory 0.01 (0.02) Rebel Party*Rebel Victory 0.02 (0.04) Peace Agreement −0.01 (0.02) Rebel Party*Peace Agreement 0.07 (0.04) Low Activity 0.01 (0.03) Rebel Party*Low Activity −0.15∗ (0.06) Constant 0.06 0.07 0.07 (0.09) (0.09) (0.09) Country Variance 0.00 0.00 0.00 AIC −403.13 −405.06 −410.42 Log Likelihood 217.57 218.53 221.21 Num. obs. 811 811 811 ∗ ∗∗ ∗∗∗ p<0.1; p<0.05; p<0.01 Time splines omitted 109 Table A5: Multilevel Linear Regression Model (Ongoing Violence) % Women Candidates % Women Elected (11) (12) (13) ∗∗ ∗∗∗ Rebel Party 0.07 0.08 0.10∗∗∗ (0.03) (0.02) (0.02) Democracy −0.07 −0.03 −0.03 (0.10) (0.01) (0.01) Gender Quotas −0.08 0.04∗ 0.04∗ (0.07) (0.02) (0.02) ∗ Candidate Selection −0.02 −0.04 −0.04∗ (0.06) (0.02) (0.02) Opposition Party −0.02 0.01 0.02 (0.03) (0.02) (0.02) Party Institutionalization 0.05 0.28∗∗∗ 0.28∗∗∗ (0.23) (0.07) (0.07) log(GDP per Capita) 0.04 −0.04∗∗ −0.04∗∗ (0.05) (0.01) (0.01) Women’s Civil Society Part. 0.24 0.13∗ 0.13∗ (0.31) (0.06) (0.06) ∗ Ongoing Conflict 0.06 −0.04 −0.03 (0.06) (0.02) (0.02) Rebel Party*Ongoing Conflict −0.05 (0.04) Constant −0.24 0.19 0.18 (0.34) (0.10) (0.10) Country Variance 0.01 0.00 0.00 AIC −104.07 −408.78 −404.11 Log Likelihood 67.03 219.39 218.06 No. Obs. 124 808 808 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Time splines omitted 110 Table A6: Multilevel Linear Regression Model (Women’s Participation in Rebel Groups) % Women Elected (14) Female Members 0.10∗ (0.05) Nationalist Ideology 0.04 (0.03) log(Seat Share) 0.04∗∗∗ (0.01) Democracy −0.08∗ (0.04) Gender Quotas 0.03 (0.03) Candidate Selection 0.03 (0.03) Opposition Party 0.01 (0.03) Party Institutionalization 0.43∗∗∗ (0.12) log(GDP per Capita) −0.03 (0.02) Women’s Civil Society Participation 0.25∗ (0.11) Constant −0.11 (0.16) Country Variance 0.00 (0.01) N. Obs. 91 AIC −99.08 Log Likelihood 65.54 ∗ ∗∗ ∗∗∗ p<0.1; p<0.05; p<0.01 Time splines omitted 111 Table A7: Multilevel Linear Regression Model (Alternate Democracy Operationalization) % Women Candidates % Women Elected (15) (16) (17) (18) (19) (20) Rebel Party 0.07∗ 0.07∗∗ 0.07∗∗ 0.07∗∗ 0.08∗∗∗ 0.08∗∗∗ (0.03) (0.03) (0.03) (0.02) (0.02) (0.02) Democracy −0.05 −0.03∗ (0.10) (0.02) Gender Quotas −0.05 −0.05 −0.06 0.03∗ 0.02 0.02 (0.06) (0.05) (0.05) (0.02) (0.02) (0.02) Candidate Selection −0.02 −0.01 −0.01 −0.03 −0.04∗ −0.04∗ (0.06) (0.05) (0.05) (0.02) (0.02) (0.02) Opposition Party −0.02 −0.02 −0.02 0.02 0.02 0.02 (0.03) (0.03) (0.03) (0.02) (0.02) (0.02) Party Institutionalization 0.10 0.07 0.09 0.24∗∗∗ 0.25∗∗∗ 0.25∗∗∗ (0.23) (0.21) (0.20) (0.07) (0.07) (0.07) log(GDP per Capita) 0.04 0.06 0.06 −0.03∗ −0.02 −0.02 (0.05) (0.05) (0.04) (0.01) (0.01) (0.01) Women’s Civil Society Part. 0.14 0.32 0.44 0.21∗∗∗ 0.28∗∗∗ 0.29∗∗∗ (0.29) (0.28) (0.29) (0.05) (0.05) (0.06) Rebel Party*Democracy −0.00 0.05 (0.05) (0.04) Electoral Democracy (V-Dem) −0.32 −0.21∗∗ (0.23) (0.07) Liberal Democracy (V-Dem) −0.42 −0.20∗ (0.25) (0.09) Constant −0.14 −0.32 −0.43 0.06 0.00 −0.04 (0.32) (0.32) (0.32) (0.09) (0.08) (0.09) Country Variance 0.01 0.01 0.01 0.00 0.00 0.00 AIC −102.67 −112.05 −112.91 −411.83 −426.12 −421.63 Log Likelihood 66.33 70.02 70.45 220.91 227.06 224.81 No. Obs. 124 124 124 811 811 811 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Time splines omitted 112 Table A8: Multilevel Linear Regression Model (Split Sample) % Women Elected Autocracies Democracies (21) (22) ∗ Rebel Party 0.06 0.12∗∗∗ (0.02) (0.03) ∗ Gender Quotas 0.08 −0.04 (0.03) (0.03) Candidate Selection −0.06∗ 0.02 (0.03) (0.03) Opposition Party 0.00 0.03 (0.03) (0.03) ∗ Party Institutionalization 0.26 0.00 (0.13) (0.16) log(GDP per Capita) −0.03 −0.00 (0.02) (0.03) ∗ Women’s Civil Society Part. 0.22 0.27∗∗∗ (0.09) (0.08) Constant 0.09 −0.09 (0.17) (0.21) Country Variance 0.00 0.00 AIC −150.82 −225.69 BIC −99.03 −173.35 Log Likelihood 88.41 125.84 Num. obs. 397 414 ∗ ∗∗ ∗∗∗ p<0.1; p<0.05; p<0.01 Time splines omitted 113 Table A9: Multilevel Linear Regression Model (Alternate Gender Quotas Operationalization) % Women Candidates % Women Elected (23) (24) (25) (26) (27) (28) Rebel Party 0.07∗∗ 0.07∗∗ 0.06 0.08∗∗∗ 0.08∗∗∗ 0.05∗ (0.03) (0.03) (0.03) (0.02) (0.02) (0.02) Democracy −0.05 −0.06 −0.02 −0.03∗ −0.00 −0.03 (0.10) (0.11) (0.09) (0.01) (0.02) (0.02) Implemented Quotas −0.05 0.03 (0.06) (0.02) Effective Quotas −0.04 0.09∗∗∗ (0.08) (0.02) Party Quota 0.05 0.12∗ (0.05) (0.06) Rebel Party*Party Quota −0.04 0.05 (0.06) (0.07) Candidate Selection −0.02 −0.03 −0.02 −0.04∗ −0.02 −0.03 (0.05) (0.05) (0.05) (0.02) (0.02) (0.02) Opposition Party −0.02 −0.02 −0.02 0.02 0.02 0.02 (0.03) (0.03) (0.03) (0.02) (0.02) (0.02) Party Institutionalization 0.09 0.19 0.13 0.24∗∗∗ 0.12 0.14 (0.23) (0.24) (0.22) (0.07) (0.07) (0.08) log(GDP per Capita) 0.04 0.02 0.03 −0.03∗ −0.01 −0.02 (0.05) (0.05) (0.05) (0.01) (0.01) (0.02) Women’s Civil Society Part. 0.14 0.14 0.09 0.20∗∗∗ 0.19∗∗∗ 0.17∗∗ (0.29) (0.31) (0.27) (0.05) (0.05) (0.06) Constant −0.14 −0.09 −0.16 0.06 −0.04 0.05 (0.32) (0.33) (0.32) (0.08) (0.09) (0.11) Country Variance 0.01 0.01 0.01 0.00 0.00 0.00 AIC −108.77 −108.86 −102.58 −415.87 −426.14 −428.75 Log Likelihood 68.39 68.43 66.29 221.94 227.07 229.37 No. Obs. 124 124 124 811 811 811 ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 Time splines omitted 114 Table A10: Multilevel Linear Regression Results (Seat Share) % Women Elected (29) Rebel Party 0.05∗ (0.02) log(Seat Share) 0.01∗∗ (0.00) log(GDP per Capita) −0.03∗ (0.01) Level of Democracy −0.18∗∗ (0.07) Gender Quotas 0.02 (0.02) Candidate Selection −0.03 (0.02) Opposition Party 0.00 (0.02) Party Institutionalization 0.28∗∗∗ (0.07) Women’s Civil Society Part. 0.25∗∗∗ (0.05) Constant 0.09 (0.09) Country Variance 0.00 (0.03) AIC −416.28 Log Likelihood 223.14 Num. Obs. 805 ∗ ∗∗ ∗∗∗ p<0.1; p<0.05; p<0.01 Time splines omitted 115 Table A11: Tobit Model (Alternate Specification) % Women Elected (30) Rebel Party 0.32∗∗∗ (0.05) Democracy −0.08 (0.04) GenderQuotas 0.01 (0.05) Candidate Selection −0.18∗∗∗ (0.05) Opposition Party 0.14∗∗ (0.05) Party Institutionalization 0.29 (0.19) log(GDP per Capita) −0.05 (0.04) Women’s Civil Society Participation 0.42∗∗ (0.14) Constant −0.23 (0.24) Sigma −0.88∗∗∗ (0.05) AIC 749.97 Log Likelihood −361.99 No. Obs. 811 ∗ ∗∗ ∗∗∗ p<0.1; p<0.05; p<0.01 Time splines omitted 116 APPENDIX B APPENDIX TO CHAPTER 2 Table B1: Descriptive Statistics Statistic N Mean St. Dev. Min Pctl(25) Pctl(75) Max Proportion of Women’s Seats 261 0.17 0.17 0 0 0.32 0.57 Female Combatants 254 0.57 0.49 0 0 1 1 Level of Democracy 302 0.42 0.19 0.07 0.26 0.55 0.88 Gender Quotas 301 0.29 0.46 0 0 1 1 log(GDP per Capita) 285 8.20 1.01 6.24 7.35 8.95 10.53 Fem. Labor Force Part. 279 53.82 20.03 8.54 37.19 69.78 87.39 Fertility Level 286 4.01 1.71 1.27 2.46 5.45 7.46 Leftist Ideology 297 0.19 0.39 0 0 0 1 Civilian Support 233 0.75 0.43 0 1 1 1 External Support 264 0.74 0.44 0 0 1 1 117 Table B2: Tobit Model (Voter Perceptions) Dependent Variable: % Women’s Seats by Party (1) (2) (3) (4) (5) (6) ∗∗ ∗ Female Combatants 0.176 0.187 0.230 0.114 0.169 0.199∗∗ (0.057) (0.079) (0.141) (0.068) (0.189) (0.065) Democracy -0.286 -0.286 -0.265 -0.290 -0.262 -0.0992 (0.163) (0.165) (0.186) (0.164) (0.186) (0.219) log(GDP per Capita) 0.034 0.034 0.045 0.044 0.060 -0.052 (0.042) (0.042) (0.054) (0.042) (0.052) (0.073) Gender Quotas 0.169∗∗∗ 0.170∗∗∗ 0.167∗∗∗ 0.174∗∗∗ 0.172∗∗∗ 0.266∗∗∗ (0.041) (0.042) (0.040) (0.042) (0.041) (0.067) Female Labor Force Part. 0.003 0.003 0.003 0.003∗ 0.004 0.013∗∗∗ (0.002) (0.002) (0.002) (0.002) (0.002) (0.004) Fertility Levels -0.034 -0.035 -0.029 -0.030 -0.022 -0.038 (0.022) (0.023) (0.026) (0.021) (0.025) (0.046) Leftist 0.062 0.089 0.064 0.072 0.059 0.709∗∗∗ (0.064) (0.124) (0.065) (0.062) (0.133) (0.120) Rebel Support 0.215∗∗∗ 0.216∗∗∗ 0.210∗∗∗ 0.167∗ 0.151∗ 0.630∗∗∗ (0.048) (0.048) (0.052) (0.073) (0.074) (0.108) Civilian Aid 0.236∗∗∗ 0.236∗∗∗ 0.271∗ 0.239∗∗∗ 0.286∗∗ 0.416∗ (0.055) (0.055) (0.105) (0.054) (0.107) (0.190) Women’s Exclusion 0.016 0.031 0.0356 0.015 0.032 -0.688 (0.142) (0.175) (0.163) (0.142) (0.198) (0.494) Women’s Property Rights 0.025 0.024 0.023 0.025 0.023 -0.025 (0.032) (0.033) (0.034) (0.032) (0.034) (0.052) log(ODA) 0.011 0.009 0.009 0.015 0.015 0.026 (0.019) (0.020) (0.019) (0.019) (0.020) (0.034) Women’s Civil Society Part 0.031 0.029 0.028 0.029 0.026 0.006 (0.032) (0.034) (0.033) (0.0334) (0.035) (0.044) Female Combat*Leftist -0.034 0.022 -0.584∗∗∗ (0.136) (0.150) (0.121) Female Combat*Civilian Aid -0.058 -0.077 0.637 (0.164) (0.161) Female Combat*External Support 0.087 0.102 0.315∗ (0.091) (0.099) (0.134) Constant -0.835 -0.810 -0.969 -1.017 -1.244 -0.303 (0.676) (0.671) (0.841) (0.681) (0.787) (1.151) Sigma 0.020∗∗∗ 0.020∗∗∗ 0.020∗∗∗ 0.020∗∗∗ 0.020∗∗∗ 0.010∗∗ (0.005) (0.005) (0.005) (0.004) (0.005) (0.004) Country Fixed Effects N N N N N Y No. Obs. 130 130 130 130 130 130 Pseudo R2 2.045 2.047 2.049 2.066 2.073 3.415 Log Likelihood 29.06 29.12 29.17 29.65 29.86 67.19 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 118 Table B3: Tobit Model (Seat Share) Dependent Variable: % Women’s Seats by Party (7) (8) (9) (10) Female Combatants 0.209∗∗∗ 0.068 0.259∗∗∗ 0.265∗∗∗ (0.054) (0.042) (0.047) (0.051) Democracy -0.207 -0.169 -0.126 -0.132 (0.127) (0.120) (0.151) (0.147) log(GDP per Capita) 0.014 0.013 0.014 0.015 (0.022) (0.019) (0.033) (0.0332) Gender Quotas 0.175∗∗∗ 0.192∗∗∗ 0.149∗∗∗ 0.152∗∗∗ (0.039) (0.037) (0.039) (0.042) Female Labor Force Participation 0.003∗ 0.002 0.002 0.002 (0.001) (0.001) (0.001) (0.001) Fertility Levels -0.043∗ -0.043∗∗ -0.033 -0.033 (0.018) (0.015) (0.019) (0.019) Leftist Ideology 0.065 0.103∗ 0.105∗ 0.104∗ (0.060) (0.043) (0.049) (0.047) External Support 0.207∗∗∗ 0.195∗∗∗ 0.109∗ 0.118 (0.049) (0.041) (0.050) (0.062) Civilian Aid 0.217∗∗∗ 0.152∗∗∗ 0.171∗ 0.172∗ (0.052) (0.040) (0.069) (0.067) Seat Share 0.001 0.002∗ (0.001) (0.001) Female Combatants*Seat Share 0.308∗∗∗ (0.062) Opposition Party -0.104∗∗ -0.081 (0.034) (0.057) Female Combatants*Opposition Party -0.026 (0.078) Constant -0.393 -0.285 -0.294 -0.319 (0.258) (0.181) (0.321) (0.332) Sigma 0.021∗∗∗ 0.016∗∗∗ 0.016∗∗∗ 0.016∗∗∗ (0.004) (0.004) (0.004) (0.004) No. Obs. 138 138 111 111 Pseudo R2 1.915 2.377 2.550 2.553 Log Likelihood 27.19 40.90 36.10 36.15 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 119 Table B4: Tobit Model (Alternate Operationalizations) Dependent Variable: % Women’s Seats by Party (11) (12) (13) (14) (15) Female Combatants 0.218∗∗∗ 0.225∗∗∗ 0.201∗∗∗ 0.186∗∗ (0.061) (0.056) (0.056) (0.062) Prop. Female Combatants 0.083∗∗∗ (0.024) Democracy (V-Dem) -0.079 -0.241 -0.218 -0.243 (0.108) (0.132) (0.128) (0.133) Polity2 -0.010 (0.006) log(GDP per Capita) 0.022 -0.017 0.019 0.008 (0.024) (0.029) (0.023) (0.025) log(GDP per Capita) (WB) 0.026 (0.018) Adopted Gender Quotas 0.196∗∗∗ 0.169∗∗∗ 0.166∗∗∗ (0.038) (0.034) (0.037) Implemented Gender Quotas 0.169∗∗∗ (0.044) Effective Gender Quotas 0.129∗ (0.056) Female Labor Force Part. 0.003∗∗ 0.002 0.003∗∗ 0.003∗∗ 0.002 (0.001) (0.001) (0.001) (0.001) (0.001) Fertility Levels -0.037∗ -0.055∗ -0.036∗ -0.041∗ -0.037 (0.018) (0.024) (0.015) (0.019) (0.025) Leftist Ideology -0.002 0.060 0.085 0.079 0.058 (0.076) (0.062) (0.061) (0.061) (0.065) External Support 0.215∗∗∗ 0.202∗∗∗ 0.227∗∗∗ 0.217∗∗∗ 0.194∗∗ (0.051) (0.046) (0.054) (0.050) (0.061) Civilian Support 0.192∗∗∗ 0.199∗∗∗ 0.213∗∗∗ 0.207∗∗∗ 0.162∗∗ (0.054) (0.053) (0.054) (0.053) (0.052) Constant -0.496∗ -0.078 -0.542∗∗ -0.431 -0.203 (0.233) (0.389) (0.191) (0.284) (0.331) Sigma 0.021∗∗∗ 0.018∗∗∗ 0.021∗∗∗ 0.022∗∗∗ 0.024∗∗∗ (0.004) (0.004) (0.005) (0.004) (0.005) N. Obs. 138 130 134 138 138 Pseudo R2 1.866 2.266 1.776 1.842 1.646 Log Likelihood 25.72 33.09 24.74 25.00 19.18 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 120 Table B5: Multiple Regression Analysis (Alternate Specification) Dependent Variable: % Women’s Seats by Party (16) (17) Female Combatants 0.152∗∗∗ 0.203∗∗∗ (0.0366) (0.0535) Level of Democracy -0.279∗∗ -0.215 (0.0940) (0.121) log(GDP per Capita) 0.00753 0.0124 (0.0195) (0.0255) Gender Quotas 0.118∗∗ 0.166∗∗∗ (0.0355) (0.0477) Female Labor Force Participation 0.000997 0.00258∗ (0.00106) (0.00113) Fertility Levels -0.0286 -0.0427∗ (0.0151) (0.0171) Leftist Ideology 0.0629 0.0647 (0.0356) (0.0601) Civilian Support 0.0985∗ 0.215∗∗∗ (0.0472) (0.0548) External Support 0.103∗∗ 0.211∗∗∗ (0.0289) (0.0486) Constant 0.123 -0.384 (0.232) (0.296) Year FEs Y N Year Splines N Y Sigma 0.0208∗∗∗ (0.00459) N. Obs. 138 138 R2 0.623 Pseudo R2 1.911 Log Likelihood 117.1 27.08 Clustered Standard Errors in Parentheses Year Splines and Year Fixed Effects Omitted Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 121 Table B6: Zero Inflated Beta Regression (Alternate Specification) Proportion: Women’s Seats by Party (18) (19) (20) (21) Female Combatants 0.758∗∗∗ 0.664∗∗ 1.053 0.792∗ (0.208) (0.231) (0.566) (0.340) Democracy -0.114 0.000 -0.097 -0.114 (0.497) (0.483) (0.492) (0.497) log(GDP per Capita) -0.027 -0.020 -0.016 -0.028 (0.092) (0.091) (0.093) (0.093) Gender Quotas 0.364∗∗ 0.350∗∗ 0.387∗∗ 0.362∗∗ (0.131) (0.126) (0.134) (0.129) Female Labor Force Part. 0.012∗∗ 0.011∗∗ 0.010∗∗ 0.011∗∗ (0.004) (0.004) (0.004) (0.004) Fertility Levels -0.208∗∗ -0.196∗∗ -0.191∗∗ -0.208∗∗ (0.063) (0.064) (0.063) (0.063) Leftist Ideology 0.059 -0.515 0.083 0.058 (0.179) (0.424) (0.179) (0.180) External Support 0.267 0.183 0.299 0.285 (0.183) (0.192) (0.189) (0.206) Civilian Support 0.170 0.156 0.467 0.170 (0.175) (0.181) (0.517) (0.176) Female Combat.*Leftist 0.672 (0.460) Female Combat.*Civilian Support -0.369 (0.568) Female Combat*External Support -0.0411 (0.350) Constant -1.629 -1.619 -2.056 -1.633 (1.054) (1.004) (1.306) (1.056) Zero Inflated: No Women Elected Female Combatants -0.709 -0.931 0.449 -0.523 (0.653) (0.770) (1.020) (0.904) Democracy -1.570 -1.381 -1.777 -1.558 (1.736) (1.846) (1.701) (1.757) log(GDP per Capita) 0.260 0.256 0.383 0.249 (0.322) (0.325) (0.316) (0.328) Gender Quota -1.055 -1.076 -1.062 -1.069 (0.666) (0.667) (0.639) (0.671) Female Labor Force Part. -0.015 -0.015 -0.016 -0.015 (0.015) (0.015) (0.015) (0.015) Fertility Level 0.152 0.153 0.247 0.146 (0.212) (0.220) (0.227) (0.213) Leftist Ideology -0.358 -1.252 -0.250 -0.378 (0.826) (0.909) (0.898) (0.833) External Support -1.706∗∗ -1.793∗∗ -1.772∗∗ -1.631∗ (0.586) (0.574) (0.598) (0.678) Civilian Support -1.358 -1.415 -0.467 -1.379 (0.694) (0.702) (0.683) (0.722) Female Combat.*Leftist 1.334 (1.128) Female Combat.*Civilian Support -1.539 (1.030) Female Combat.*External Support -0.247 (1.111) Constant 1.147 1.328 -0.777 1.256 (3.192) (3.332) (3.129) (3.268) N. Obs. 261 261 261 261 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 122 Table B7: Heckman Selection Model (Alternate Specification) Proportion: Women’s Seats (22) Female Combatants 0.121∗∗ (0.055) 0.093 Democracy (0.185) log(GDP per Capita) -0.024 (0.041) Gender Quotas 0.192∗∗∗ (0.053) Female Labor Force Participation 0.000 (0.001) Fertility Levels -0.002 (0.021) Leftist Ideology 0.110∗∗ (0.053) Civilian Aid -0.019 (0.051) External Support 0.028 (0.051) Rebel-to-Party Transformation Female Combatants 0.610∗∗ (0.300) Rebel Legal Institutions 0.787∗∗ (0.395) Civilian Aid -0.179 (0.386) log(Conflict Duration) 0.183 (0.190) log(Troop Size)∗ 0.221 (0.118) UN Peacekeepers 0.765 (0.495) Electoral Participation Provision 1.12∗∗ (0.539) Conflict Settlement Agreement 0.640∗ (0.361) Territorial Control -0.419 (0.432) Cold War 0.880∗∗ (0.431) Mills Ratio -0.052 Sigma -0.113 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 123 Table B8: Tobit Model (Conflict Characteristics) Dependent Variable: % Women’s Seats by Party (23) (24) (25) (26) (27) ∗∗ ∗∗ ∗∗∗ ∗∗∗ Female Combatants 0.174 0.197 0.214 0.211 0.215∗∗∗ (0.056) (0.064) (0.055) (0.053) (0.055) Democracy -0.214 -0.192 -0.141 -0.214 -0.235∗ (0.143) (0.142) (0.143) (0.119) (0.116) log(GDP per Capita) 0.006 0.014 0.019 0.066∗ -0.005 (0.021) (0.028) (0.020) (0.029) (0.026) Gender Quotas 0.193∗∗∗ 0.188∗∗∗ 0.192∗∗∗ 0.167∗∗∗ 0.191∗∗∗ (0.053) (0.041) (0.040) (0.035) (0.042) Female Labor Force Part. 0.002∗ 0.003∗ 0.002∗ 0.003∗∗∗ 0.003∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) Fertility Levels -0.039 -0.041 -0.031 -0.032∗ -0.066∗∗ (0.025) (0.022) (0.018) (0.016) (0.024) Leftist Ideology 0.095 0.050 0.070 0.061 0.070 (0.060) (0.064) (0.061) (0.058) (0.051) External Support 0.215∗∗∗ 0.218∗∗∗ 0.197∗∗∗ 0.219∗∗∗ 0.229∗∗∗ (0.053) (0.053) (0.049) (0.043) (0.050) Civilian Support 0.194∗∗∗ 0.231∗∗∗ 0.218∗∗∗ 0.277∗∗∗ 0.205∗∗∗ (0.047) (0.050) (0.056) (0.060) (0.051) Forced Recruitment 0.029 (0.044) Troop Size 0.000 (0.000) Islamist Ideology -0.202∗∗∗ (0.054) Nationalist Ideology -0.005 (0.036) Secular Non-Left Ideology -0.088 (0.085) Conflict Duration -0.006∗∗ (0.002) Peacekeepers -0.077 (0.053) Constant -0.321 -0.434 -0.487∗ -0.878∗∗ -0.154 (0.255) (0.383) (0.231) (0.301) (0.292) Sigma 0.021∗∗∗ 0.021∗∗∗ 0.021∗∗∗ 0.0194∗∗∗ 0.020∗∗∗ (0.005) (0.005) (0.004) (0.004) (0.004) N. Obs. 118 124 132 138 137 Pseudo R2 1.937 3.192 1.767 2.041 2.091 Log Likelihood 23.57 30.64 25.03 30.94 30.84 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 124 Table B9: Tobit Model (Rebel Governance) Dependent Variable: % Women’s Seats by Party (28) (29) (30) (31) (32) (33) (34) ∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗ Female Combatants 0.227 0.211 0.228 0.228 0.216 0.208 0.236∗∗∗ (0.057) (0.054) (0.055) (0.056) (0.054) (0.055) (0.056) Democracy -0.259∗ -0.222 -0.253∗ -0.271∗ -0.233 -0.219 -0.280∗ (0.131) (0.129) (0.120) (0.123) (0.132) (0.129) (0.128) log(GDP per Capita) 0.035 0.012 0.005 0.008 0.016 0.015 0.005 (0.021) (0.022) (0.024) (0.021) (0.021) (0.022) (0.033) Gender Quotas 0.168∗∗∗ 0.172∗∗∗ 0.165∗∗∗ 0.141∗∗∗ 0.166∗∗∗ 0.173∗∗∗ 0.144∗∗∗ (0.038) (0.040) (0.040) (0.042) (0.040) (0.039) (0.042) Female Labor Force Part. 0.003∗∗ 0.003∗∗ 0.003∗∗ 0.002∗ 0.003∗∗ 0.003∗ 0.003∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Fertility -0.047∗∗ -0.046∗∗ -0.058∗∗ -0.045∗∗ -0.046∗ -0.043∗ -0.054∗∗ (0.017) (0.017) (0.020) (0.016) (0.018) (0.018) (0.017) Leftist 0.057 0.069 0.075 0.035 0.060 0.064 0.063 (0.053) (0.064) (0.053) (0.052) (0.056) (0.057) (0.058) External Support 0.222∗∗∗ 0.214∗∗∗ 0.238∗∗∗ 0.189∗∗∗ 0.209∗∗∗ 0.210∗∗∗ 0.225∗∗∗ (0.046) (0.047) (0.054) (0.048) (0.046) (0.049) (0.055) Civilian Support 0.257∗∗∗ 0.227∗∗∗ 0.241∗∗∗ 0.220∗∗∗ 0.225∗∗∗ 0.221∗∗∗ 0.225∗∗∗ (0.049) (0.050) (0.047) (0.049) (0.051) (0.054) (0.059) Rebel Legal Institutions -0.072 -0.035 (0.037) (0.044) Rebel Elections -0.029 -0.014 (0.054) (0.069) Rebel Police -0.077 -0.061 (0.044) (0.051) Rebel Humanitarian Aid -0.144∗ -0.101 (0.073) (0.094) Rebel Education -0.030 0.004 (0.032) (0.033) Rebel Diplomacy -0.003 0.047 (0.043) (0.043) Constant -0.586∗ -0.381 -0.272 -0.251 -0.380 -0.406 -0.228 (0.254) (0.264) (0.256) (0.250) (0.249) (0.264) (0.313) Sigma 0.020∗∗∗ 0.021∗∗∗ 0.020∗∗∗ 0.020∗∗∗ 0.021∗∗∗ 0.021∗∗∗ 0.019∗∗∗ (0.005) (0.005) (0.005) (0.004) (0.005) (0.004) (0.004) No. Obs. 138 138 138 138 138 138 138 Pseudo R2 1.979 1.916 1.998 2.017 1.922 1.905 2.095 Log Likelihood 29.09 27.22 29.64 30.20 27.39 26.88 32.55 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 125 Table B10: Tobit Model (Time) Dependent Variable: % Women’s Seats by Party (35) (36) ∗∗∗ Female Combatants 0.202 0.193∗∗ (0.049) (0.071) Democracy -0.146 -0.156 (0.126) (0.135) log(GDP per Capita) 0.002 0.003 (0.019) (0.019) Gender Quotas 0.162∗∗∗ 0.161∗∗∗ (0.035) (0.036) Female Labor Force Participation 0.002∗ 0.002∗ (0.001) (0.001) Fertility Levels -0.034∗ -0.035∗ (0.016) (0.016) Leftist Ideology 0.065 0.064 (0.052) (0.052) External Support 0.202∗∗∗ 0.202∗∗∗ (0.042) (0.042) Civilian Support 0.197∗∗∗ 0.198∗∗∗ (0.049) (0.048) Time 0.005∗∗ 0.004 (0.002) (0.002) Female Combatants*Time 0.001 (0.004) Constant -0.358 -0.352 (0.205) (0.211) Sigma 0.019∗∗∗ 0.019∗∗∗ (0.005) (0.005) N. Obs. 138 138 Pseudo R2 2.055 2.056 Log Likelihood 31.34 31.36 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 126 Table B11: Tobit Model (Conflict Outcome) Dependent Variable: % Women’s Seats by Party (37) (38) (39) (40) (41) (42) ∗∗∗ ∗ ∗∗∗ ∗∗∗ ∗∗∗ Female Combatants 0.203 0.199 0.205 0.205 0.183 0.192∗∗∗ (0.047) (0.059) (0.059) (0.053) (0.047) (0.055) Democracy -0.074 -0.077 -0.247 -0.220 -0.174 -0.179 (0.143) (0.146) (0.149) (0.126) (0.112) (0.122) log(GDP per Capita) 0.012 0.013 0.019 0.018 0.036 0.033 (0.021) (0.019) (0.024) (0.022) (0.020) (0.028) Gender Quotas 0.203∗∗∗ 0.202∗∗∗ 0.165∗∗∗ 0.173∗∗∗ 0.156∗∗∗ 0.160∗∗∗ (0.041) (0.041) (0.043) (0.039) (0.038) (0.041) Female Labor Force Part. 0.003∗∗ 0.003∗∗ 0.002∗ 0.003∗∗ 0.003∗∗ 0.003∗∗ (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) Fertility Levels -0.045∗∗ -0.045∗∗ -0.039∗ -0.041∗ -0.036∗ -0.031 (0.015) (0.015) (0.017) (0.018) (0.017) (0.020) Leftist Ideology 0.027 0.028 0.062 0.065 0.030 0.078 (0.049) (0.050) (0.062) (0.060) (0.052) (0.060) External Support 0.204∗∗∗ 0.205∗∗∗ 0.207∗∗∗ 0.204∗∗∗ 0.202∗∗∗ 0.198∗∗∗ (0.038) (0.037) (0.051) (0.049) (0.046) (0.050) Civilian Support 0.248∗∗∗ 0.248∗∗∗ 0.214∗∗∗ 0.213∗∗∗ 0.195∗∗∗ 0.186∗∗∗ (0.044) (0.045) (0.051) (0.051) (0.053) (0.050) Rebel Victory 0.126∗∗ 0.119∗ (0.057) (0.037) Female Combatants*Rebel Victory 0.010 (0.064) Peace Agreement 0.017 (0.044) Cease Fire -0.618∗∗∗ (0.083) Low Activity -0.177∗∗∗ (0.048) Remobilization -0.067 (0.056) Constant -0.478 -0.483∗ -0.418 -0.424 -0.527 -0.574 (0.241) (0.226) (0.259) (0.265) (0.271) (0.302) Sigma 0.019∗∗∗ 0.019∗∗∗ 0.021∗∗∗ 0.021∗∗∗ 0.018∗∗∗ 0.020∗∗∗ (0.004) (0.004) (0.005) (0.004) (0.004) (0.004) N. Obs. 133 133 133 138 138 138 Pseudo R2 2.182 2.182 1.939 1.919 2.226 1.948 Log Likelihood 32.55 32.56 25.86 27.30 36.42 28.15 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 127 Figure A1: Marginal Effect of Female Combatants and Time Figure A2: Marginal Effect of Female Combatants and Rebel Victory 128 APPENDIX C APPENDIX TO CHAPTER 3 To code the election of former female rebels, I first collected a list of all women who were elected to the sample of rebel parties for a given election. To identify such lists, I relied on parliamentary records, election results and reports, and media reporting on elections. In some cases, MPs were directly listed as males or females. In other cases, names were accompanied by pictures of the MP, which also allowed for a rudimentary identification of gender. If neither of the prior information was available, further background information was conducted on the individual MP to determine their gender. After collecting a list of the female MPs elected, background research was conducted on each MP to determine if they were or were not former members of the respective rebel groups. To determine rebel status, I relied on a variety of sources, including parliamentary profiles, media interviews, policy reports, academic research, and biographical profiles. To code a MP as a former rebel, their participation had to be clearly specified. They had to be identified as a former member, soldier, cadre, or leader during the time period in which the group operated as a non-state armed group. These conditions require that knowledge of former rebel status is public and known. While in many of these contexts, given that the prominence of these groups in politics, MPs are unlikely to hide or cover up their rebel backgrounds, there may be MPs who do so. As a result, through this coding, the number of former female rebels may be underestimated. Of the 1,956 women coded in the dataset, approximately 20 percent were coded as missing. In addition to coding whether MPs were members of the respective rebel groups, the dataset also includes whether MPs were rebel leaders or had familial or romantic connections 129 to rebels, as well as incumbency status, education levels, professional background, and age. Table C1: Descriptive Statistics Statistic N Mean St. Dev. Min Pctl(25) Pctl(75) Max Prop. Female Rebels Elected 155 0.17 0.26 0 0 0.26 1 Female Combatants 161 0.81 0.40 0 1 1 1 Prop. Female Combatants 160 1.41 0.99 0 1 2 3 Female Leaders 171 0.72 0.45 0 0 1 1 Women’s Access to State Jobs 192 0.81 0.95 −2.07 −0.17 1.50 2.47 Women’s Civil Society Part. 200 1.37 0.69 −2.64 0.98 1.84 2.68 Level of Democracy 200 0.46 0.19 0.10 0.29 0.60 0.88 log(GDP per Capita) 187 8.32 0.99 6.24 7.49 9.01 10.53 Gender Quotas 192 0.33 0.47 0 0 1 1 Rebel Victory 185 0.20 0.40 0 0 0 1 Fertility Levels 180 3.69 1.71 1.27 2.25 5.33 7.14 130 Table C2: Tobit Model (Conflict Factors) Dependent Variable: % Female Rebels Elected (1) (2) (3) (4) (5) (6) ∗∗∗ ∗∗∗ ∗∗ ∗∗∗ ∗∗∗ Proportion of Female Combatants 0.512 0.428 0.247 0.329 0.306 0.262∗∗∗ (0.109) (0.081) (0.080) (0.077) (0.056) (0.058) Forced Recruitment -0.233 (0.285) Democracy -0.052 0.108 0.155 -0.063 0.219 -0.056 (0.459) (0.393) (0.435) (0.406) (0.332) (0.512) log(GDP per Capita) 0.239 0.302∗ 0.406∗∗ 0.518∗∗ 0.239 0.243∗ (0.167) (0.139) (0.142) (0.174) (0.130) (0.113) Gender Quotas -0.108 0.0984 0.232 0.217 0.154 0.079 (0.175) (0.119) (0.184) (0.174) (0.136) (0.135) Rebel Victory -0.055 -0.043 0.480∗ 0.674∗∗ -0.039 -0.053 (0.241) (0.187) (0.230) (0.217) (0.187) (0.156) Fertility Levels 0.174 0.157 0.165 0.189 0.156 0.094 (0.116) (0.091) (0.116) (0.130) (0.083) (0.077) Women’s Access to State Jobs 0.170 0.005 -0.149 -0.126 -0.027 -0.038 (0.207) (0.095) (0.091) (0.082) (0.089) (0.081) Women’s Civil Society Part. -0.175 -0.303∗∗ -0.342∗∗ -0.278∗∗ -0.208 -0.223∗∗ (0.169) (0.108) (0.125) (0.084) (0.141) (0.073) Political Wing -0.179 (0.142) Rebel Governance 0.090∗∗ (0.030) Rebel Diplomacy 0.594∗∗∗ (0.166) Nationalist Ideology 0.136 (0.248) Leftist Ideology 0.298 (0.202) Islamist Ideology -1.158∗∗ (0.401) Civilian Support 0.459∗ (0.225) Constant -3.284 -3.349∗ -4.530∗∗ -5.609∗∗ -3.057∗ -2.759∗ (1.818) (1.529) (1.714) (2.025) (1.311) (1.326) Sigma 0.162 0.131∗ 0.110∗ 0.117∗ 0.119∗ 0.113∗ (0.082) (0.054) (0.048) (0.0500) (0.051) (0.048) No. Obs. 100 110 95 95 110 79 Pseudo R2 0.473 0.474 0.620 0.635 0.488 0.478 Log Likelihood -34.94 -39.42 -25.45 -24.45 -38.40 -30.31 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 131 Table C3: Tobit Model (Post-Conflict Factors) Dependent Variable: % Female Rebels Elected (7) (8) (9) (10) Proportion Female Combatants 0.359∗∗∗ 0.397∗∗∗ 0.409∗∗∗ 0.353∗∗∗ (0.096) (0.069) (0.074) (0.066) Women’s Access to State Jobs -0.088 0.076 0.128 -0.046 (0.121) (0.106) (0.098) (0.082) Women’s Civil Society Part. -0.262∗ -0.334∗ -0.385∗∗∗ -0.208 (0.117) (0.137) (0.087) (0.115) Democracy 0.227 0.106 -0.094 0.092 (0.406) (0.400) (0.311) (0.394) log(GDP per Capita) 0.484∗∗ 0.264 0.329∗ 0.383∗ (0.181) (0.150) (0.127) (0.156) Gender Quotas 0.093 0.145 0.044 0.186 (0.178) (0.138) (0.109) (0.131) Rebel Victory 0.283 -0.144 0.086 (0.173) (0.194) (0.181) Fertility Levels 0.196 0.167 0.209∗ 0.237∗ (0.148) (0.102) (0.097) (0.118) UN Peacekeepers 0.293 (0.193) Electoral Participation Provision 0.143 (0.138) Peace Agreement 0.457∗∗ (0.173) Time -0.042 (0.021) Constant -5.272∗ -3.126 -3.924∗ -4.166∗ (2.317) (1.651) (1.558) (1.733) Sigma 0.125∗ 0.128∗ 0.105∗ 0.117∗ (0.060) (0.053) (0.043) (0.046) No. Obs. 83 109 110 110 Pseudo RR2 0.524 0.477 0.570 0.500 Log Likelihood -29.25 -38.98 -32.24 -37.51 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 132 Table C4: Tobit Model (Party Strength) Dependent Variable: % Female Rebels Elected (11) (12) ∗∗∗ Proportion of Female Combatants 0.337 0.386∗∗∗ (0.0828) (0.0943) Female Leaders 0.0631 -0.231 (0.197) (0.244) Women’s Access to State Jobs -0.0418 0.0444 (0.0953) (0.121) Women’s Civil Society Participation -0.480∗∗ -0.366 (0.153) (0.208) Democracy -0.193 0.0301 (0.403) (0.361) log(GDP per Capita) 0.212 0.195 (0.166) (0.144) Gender Quotas 0.103 -0.0122 (0.129) (0.0903) Rebel Victory -0.444 -0.0678 (0.270) (0.132) Fertility Levels 0.116 0.114 (0.110) (0.0974) Seats Share 0.965∗ (0.436) Opposition Party 0.0831 (0.164) Constant -1.092 -1.096 (1.644) (1.716) Sigma 0.128∗∗ 0.0957∗∗ (0.0424) (0.0346) No. Obs 109 95 Pseudo R2 0.563 0.575 Log Likelihood -32.57 -27.95 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 133 Table C5: Tobit Analysis (Alternate DV) Dependent Variable: % Female Rebels Elected (Party Total) (13) (14) (15) (16) (17) (18) (19) ∗ ∗ ∗∗ ∗ Female Combatants 0.297 0.285 0.308 0.227 (0.115) (0.113) (0.104) (0.114) Proportion of 0.136∗∗∗ 0.118∗∗∗ Female Combatants (0.023) (0.021) Female Leaders 0.148 0.111 0.0281 (0.098) (0.070) (0.070) Women’s Access -0.111∗ -0.078∗ -0.002 to State Jobs (0.044) (0.038) (0.026) Women’s Civil -0.119∗ -0.100∗ -0.090∗ Society Participation (0.053) (0.039) (0.038) Level of Democracy 0.103 0.053 0.219 0.206 -0.016 0.167 0.033 (0.187) (0.138) (0.180) (0.154) (0.180) (0.183) (0.136) log(GDP per Capita) 0.037 0.068 -0.001 0.041 0.076 0.052 0.0791∗ (0.053) (0.042) (0.040) (0.054) (0.050) (0.047) (0.040) Gender Quotas -0.022 0.046 0.018 -0.001 -0.012 0.015 0.054 (0.067) (0.054) (0.071) (0.053) (0.049) (0.050) (0.041) Rebel Victory 0.107 0.018 0.065 0.154 0.077 0.108 -0.003 (0.108) (0.066) (0.098) (0.0888) (0.082) (0.069) (0.058) Fertility Levels 0.016 0.032 0.027 0.004 0.026 0.016 0.037 (0.036) (0.028) (0.030) (0.035) (0.032) (0.034) (0.0275) Constant -0.757 -0.975∗ -0.362 -0.703 -0.866 -0.701 -0.911∗ (0.572) (0.467) (0.420) (0.560) (0.528) (0.483) (0.402) Sigma 0.0297∗∗ 0.0152∗ 0.0243∗∗ 0.0213∗ 0.0234∗∗ 0.0170∗ 0.0119∗ (0.010) (0.006) (0.009) (0.008) (0.009) (0.007) (0.005) No. Obs. 113 112 122 113 113 112 112 Pseudo R2 0.383 1.039 0.348 0.677 0.555 0.862 1.200 Log Likelihood -19.29 1.216 -16.27 -10.10 -13.91 -4.268 6.182 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 134 Table C6: Tobit Model (Alternate Operationalizations I) Dependent Variable: % Female Rebels Elected (20) (21) (22) (23) (24) (25) Female Combatants 0.683 0.793 (0.347) (0.422) Proportion of Female Combatants 0.371∗∗∗ 0.374∗∗∗ (0.074) (0.072) Female Leaders 0.631∗∗ 0.636∗ (0.214) (0.253) Level of Democracy (Polity2) 0.023 0.006 -0.006 (0.024) (0.026) (0.029) Level of Democracy (V-Dem) 0.585 0.284 0.828 (0.666) (0.484) (0.766) log(GDP per Capita) (Maddison Project) 0.212 0.297∗ 0.182 (0.189) (0.147) (0.139) log(GDP per Capita) (WB) 0.000 0.000 -0.000 (0.000) (0.000) (0.000) Gender Quotas 0.073 0.128 0.140 -0.094 0.060 0.047 (0.165) (0.147) (0.178) (0.162) (0.148) (0.190) Rebel Victory 0.290 -0.092 -0.079 0.309 -0.110 -0.047 (0.254) (0.206) (0.223) (0.228) (0.176) (0.236) Level of Fertility 0.056 0.156 0.117 -0.009 0.051 0.147 (0.122) (0.098) (0.085) (0.092) (0.064) (0.090) Women’s Access to State Jobs -0.226 0.006 -0.120 -0.294 -0.0455 -0.111 (0.159) (0.091) (0.140) (0.164) (0.100) (0.165) Women’s CSO Participation -0.230 -0.278∗ -0.428∗∗ -0.200 -0.264 -0.382∗ (0.125) (0.112) (0.146) (0.154) (0.136) (0.154) Constant -2.318 -3.284∗ -1.610 -0.601 -0.579 -0.782 (2.000) (1.612) (1.414) (0.665) (0.517) (0.838) Sigma 0.199∗ 0.127∗ 0.190∗∗ 0.215∗ 0.140∗ 0.225∗∗ (0.088) (0.053) (0.067) (0.104) (0.065) (0.073) N. Obs. 91 90 100 109 109 119 Pseudo R2 0.247 0.436 0.223 0.257 0.417 0.201 Log Likelihood -49.17 -36.50 -57.11 -55.34 -43.42 -67.90 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 135 Table C7: Tobit Model (Alternate Observations II) Dependent Variable: % Female Rebels Elected (26) (27) (28) (29) (30) (31) ∗ ∗ Female Combatants 0.820 0.802 (0.385) (0.386) Proportion Female Combatants 0.404∗∗∗ 0.399∗∗∗ (0.0714) (0.0719) Female Leaders 0.656∗∗ 0.647∗ (0.242) (0.250) Level of Democracy 0.462 0.127 0.610 0.524 0.167 0.620 (0.477) (0.396) (0.739) (0.482) (0.410) (0.767) log(GDP per Capita) 0.224 0.308∗ 0.144 0.233 0.297∗ 0.141 (0.177) (0.149) (0.150) (0.174) (0.143) (0.146) Implemented Gender Quotas -0.0732 0.137 0.0584 (0.169) (0.152) (0.214) Effective Gender Quotas -0.00596 0.141 0.0588 (0.173) (0.141) (0.216) Rebel Victory 0.306 -0.125 -0.0284 0.316 -0.136 -0.0317 (0.234) (0.198) (0.228) (0.243) (0.198) (0.231) Level of Fertility 0.0896 0.174 0.195 0.0957 0.176 0.195 (0.117) (0.0950) (0.0992) (0.119) (0.0992) (0.101) Women’s Access to State Jobs -0.255 0.0127 -0.0900 -0.264 0.00217 -0.0924 (0.157) (0.0974) (0.166) (0.163) (0.0999) (0.169) Women’s CSO Participation -0.250 -0.289∗ -0.442∗∗ -0.250 -0.290∗ -0.441∗∗ (0.134) (0.118) (0.168) (0.131) (0.119) (0.164) Constant -2.733 -3.525∗ -1.997 -2.862 -3.430∗ -1.965 (1.924) (1.629) (1.481) (1.899) (1.604) (1.447) Sigma 0.214∗ 0.131∗ 0.221∗∗ 0.216∗ 0.133∗ 0.221∗∗ (0.0942) (0.0540) (0.0710) (0.0949) (0.0552) (0.0720) N. Obs. 111 110 120 111 110 120 Pseudo R2 0.269 0.459 0.209 0.267 0.458 0.209 Log Likelihood -55.14 -40.52 -67.65 -55.27 -40.63 -67.66 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 136 Table C8: Linear Regression Model (Alternate Specification) Dependent Variable: % Female Rebels Elected (32) (33) Proportion of Female Combatants 0.173∗∗∗ 0.178∗∗∗ (0.041) (0.042) Female Leaders 0.019 -0.011 (0.067) (0.077) Women’s Access to State Jobs -0.019 -0.031 (0.043) (0.042) Women’s Civil Society Participation -0.191∗∗∗ -0.171∗ Level of Democracy -0.108 0.191 (0.207) (0.224) log(GDP per Capita) 0.129 0.127 (0.074) (0.072) Gender Quotas 0.008 0.049 (0.052) (0.080) Rebel Victory 0.011 0.024 (0.083) (0.080) Fertility Levels 0.031 0.042 (0.040) (0.048) (0.044) (0.074) Constant -0.923 -1.067 (0.690) (0.812) No. Obs. 110 110 R2 0.518 0.676 Log Likelihood 26.93 48.84 Clustered Standard Errors in Parentheses Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01 137 REFERENCES 138 REFERENCES Acosta, Benjamin. 2014. “From bombs to ballots: When militant organizations transition to political parties.” Journal of Politics 76(3): 666–683. Alexander, Deborah and Kristi Andersen. 1993. “Gender as a Factor in the Attribution of Leadership Traits.” Political Research Quarterly 46(3): 527. Allison, Michael E. 2010. “The Legacy of Violence on Post-Civil War Elections: The Case of El Salvador.” Studies in Comparative International Development 45(1): 104–124. Annan, Jeannie, Christopher Blattman, Dyan Mazurana, and Khristopher Carlson. 2011. “Civil war, reintegration, and gender in northern Uganda.” Journal of Conflict Resolution 55(6): 877–908. Barnes, Tiffany D. and Emily Beaulieu. 2014. “Gender stereotypes and corruption: How candidates affect perceptions of election fraud.” Politics and Gender 10(3): 365–391. Barnes, Tiffany D. and Stephanie M. Burchard. 2013. ““Engendering” Politics: The Impact of Descriptive Representation on Women’s Political Engagement in Sub-Saharan Africa.” Comparative Political Studies 46(7): 767–790. Barth, Elise Friedrikke. 2002. Peace as Disappointment: The Reintegration of Female Sol- diers in Post-Conflict Societies. A Comparative Study from Africa. Technical report. Berry, Marie E. 2015a. “From violence to mobilization: Women, war, and threat in Rwanda.” Mobilization 20(2): 135–156. Berry, Marie E. 2015b. “When “bright futures” fade: Paradoxes of women’s empowerment in Rwanda.” Signs 41(1): 1–27. Berry, Marie E. 2018a. War, Women, and Power: From Violence to Mobilization in Rwanda and Bosnia-Herzegovina. Cambridge University Press. Berry, Marie E. 2018b. War, women, and power: From violence to mobilization in Rwanda and Bosnia-Herzegovina. Cambridge University Press. Berry, Marie E. and Milli Lake. 2017. “Gender, Violence and Politics in the Democratic Re- public of Congo. By Jane Freedman. Surrey: Ashgate Publishing Company, 2015. 163 pp. $109.95 (hardcover).Mobilizing Transnational Gender Politics in Post-Genocide Rwanda. By Rirhandu Mageza-Barthel. Surry:.” Politics & Gender 13(02): 336–349. Berry, Marie E. and Milli Lake. 2020. “When Quotas Come Up Short.”. Berti, Benedetta. 2016. “Rebel politics and the state: between conflict and post-conflict, resistance and co-existence.” Civil Wars 18(2): 118–136. 139 Best, Rebecca H., Sarah Shair-Rosenfield, and Reed M. Wood. 2019. “Legislative Gender Diversity and the Resolution of Civil Conflict.” Political Research Quarterly 72(1): 215– 228. Black, J. H. and L. Erickson. 2003. “Women candidates and voter bias: Do women politicians need to be better?” Electoral Studies 22(1): 81–100. Bolzendahl, Catherine and Clem Brooks. 2007. “Women’s political representation and wel- fare state spending in 12 capitalist democracies.” Social Forces 85(4): 1509–1534. Bop, C. 2001. Women in Conflicts: Theirs Gains and Their Losses. In The Aftermath. London: Zed Books pp. 19–34. Boyd, Rosalind E. 1989. “Empowerment of women in uganda: Real or symbolic.” Review of African Political Economy 16(45-46): 106–117. Braithwaite, Jessica Maves and Kathleen Gallagher Cunningham. 2020. “When Organi- zations rebel: Introducing the foundations of rebel group emergence (FORGE) dataset research note.” International Studies Quarterly 64(1): 183–193. Brechenmacher, Saskia and Caroline Hubbard. 2020. “Breaking the Cycle of Gender Ex- clusion in Political Party Development.” Carnegie Endowment for International Peace (March). Brown, Elizabeth R., Amanda B. Diekman, and Monica C. Schneider. 2011. “A change will do us good: Threats diminish typical preferences for male leaders.” Personality and Social Psychology Bulletin 37(7): 930–941. Burman, M. E. and S. McKay. 2007. “Marginalization of girl mothers during reintegration from armed groups in Sierra Leone: Original Article.” International Nursing Review 54(4): 316–323. Bush, Sarah Sunn. 2011. “International politics and the spread of quotas for women in legislatures.” International Organization 65(1): 103–137. Byanyima, W. Karagwa. 1992. Women in Political Struggle in Uganda. In Women Trans- forming Politics. pp. 129–142. Carroll, S. 2001. The impact of women in public office. Bloomington: Indiana University Press. Clayton, Amanda, Diana Z. O’Brien, and Jennifer M. Piscopo. 2019. “All Male Panels? Representation and Democratic Legitimacy.” American Journal of Political Science 63(1): 113–129. Cock, Jacklyn. 1992. Colonels and Cadres: War and Gender in South Africa. Oxford University Press. Cohen, Dara Kay. 2013a. “Explaining rape during civil war: Cross-national evidence (1980- 2009).” American Political Science Review 107(3): 461–477. 140 Cohen, Dara Kay. 2013b. “Female combatants and the perpetration of violence: Wartime rape in the Sierra Leone civil war.” World Politics 65(3): 383–415. Coppedge, Michael. 2019. “V-Dem Codebook v9.”. Coulter, Chris. 2008. “Female Fighters in the Sierra Leone War: Challenging the Assump- tions?” Feminist Review 88(1): 54–73. Cunningham, David E., Kristian Skrede Gleditsch, and Idean Salehyan. 2009. “It takes two: A dyadic analysis of civil war duration and outcome.” Journal of Conflict Resolution 53(4): 570–597. Dahal, Swechchha. 2015. Challenging the Boundaries: The Narrative of the Female Ex- Combatants in Nepal. In Female Combatants in Conflict and Peace: Challenging Gender in Violence and Post-Conflict Reintegration. Daly, Sarah Zukerman. 2019. “Voting for Victors: Why Violent Actors Win Postwar Elec- tions.” World Politics 71(4): 747–805. Daly, Sarah Zukerman. 2020. “Political life after civil wars: Introducing the Civil War Successor Party dataset.” Journal of Peace Research . de Zeeuw, Jeroen. 2008a. From Soldiers to Politicians: Transforming Rebel Movements After Civil War. Boulder, CO: Lynne Rienner. de Zeeuw, Jeroen. 2008b. Understanding the Political Transformation of Rebel Movements. Boulder, CO: Lynne Rienner. Debusscher, Petra and Maria Martin De Almagro. 2016. “Post-conflict women’s movements in turmoil: The challenges of success in Liberia in the 2005-aftermath.” Journal of Modern African Studies 54(2): 293–316. Demeritt, Jacqueline H.R., Angela D. Nichols, and Eliza G. Kelly. 2014. “Female Participa- tion and Civil War Relapse.” Civil Wars 16(3): 346–368. Deng Deng, William. 2012. Linking DDR, Security Sector Development and Transitional Justice in Southern Sudan. In Post-War Security Transitions: Participatory Peacebuilding after Asymmetric Conflicts. Denov, Myriam. 2008. “Girl Soldiers and Human Rights: Lessons from Angola, Mozambique, Sierra Leone and Northern Uganda.” The International Journal of Human Rights 12(5): 813–836. Doerrer, S., L. Kavazanjian, C. Mcconnell, and Sara Van Wie. 2011. “A gendered exami- nation of education in the reintegration strategies of Liberia and Sierra Leone.” Harvard African Policy Journal 7: 41–56. Dolan, Kathleen. 2010. “The impact of gender stereotyped evaluations on support for women candidates.” Political Behavior 32(1): 69–88. 141 Donno, Daniela and Anne Kathrin Kreft. 2019. “Authoritarian Institutions and Women’s Rights.” Comparative Political Studies 52(5): 720–753. Eckstein, Harry. 1965. “On the Etiology of Internal Wars Author ( s ).” History and Theory 4(2): 133–163. Farr, Vanessa. 2002. Gender Demilitarization as a Peacebuilding Tool. Technical Report Bonn International Center for Conversion. Findley, Michael G. and Joseph K. Young. 2007. “Fighting Fire with Fire? How (Not) to Neutralize an Insurgency.” Civil Wars 9(4): 378–401. Frederick, Brian and Matthew J. Streb. 2008. “Women running for judge: The impact of sex on candidate success in state intermediate appellate court elections.” Social Science Quarterly 89(4): 937–954. Gilmartin, Niall. 2015. “’Negotiating new roles’: Irish republican women and the politics of conflict transformation.” International Feminist Journal of Politics 17(1): 58–76. Giri, Keshab. 2021. “Do All Women Combatants Experience War and Peace Uniformly? Intersectionality and Women Combatants.” Global Studies Quarterly pp. 1–11. Goetz, Anne Marie. 2002. “No shortcuts to power: Constraints on women’s political effec- tiveness in Uganda.” Journal of Modern African Studies 40(4): 549–575. Goetz, Anne Marie. 2009. “Political Cleaners: Women as the New Anti-Corruption Force?” Gender Myths and Feminist Fables: The Struggle for Interpretive Power in Gender and Development 38(1): 85–103. Goetz, Anne Marie and Shireen Hassim. 2005. “In and Against the Party: Women’s Rep- resentation and Constituency-Building In Uganda and South Africa.” Gender Justice, Development, and Rights pp. 306–344. Goredema, Dorothy and Percyslage Chigora. 2011. “Real political empowerment or political Gimmick ? An experience of women with guns under ZANU PF during the liberation struggle of 1972-1980.” 3(March): 27–31. Gowrinathan, Nimmi. 2021. Radicalizing Her: Why Women Choose Violence. Boston: Beacon Press. Hadzic, Dino and Margit Tavits. 2019. “The gendered effects of violence on political engage- ment.” Journal of Politics 81(2): 676–680. Handrahan, Lori. 2004. “Conflict, gender, ethnicity and post-conflict reconstruction.” Secu- rity Dialogue 35(4): 429–445. Hanmer, Michael J. and Kerem Ozan Kalkan. 2013. “Behind the Curve: Clarifying the Best Approach to Calculating Predicted Probabilities and Marginal Effects from Limited Dependent Variable Models.” American Journal of Political Science 57(1): 263–277. 142 Harbom, Lotta, Erik Melander, and Peter Wallensteen. 2008. “Dyadic dimensions of armed conflict, 1946-2007.” Journal of Peace Research 45(5): 697–710. Hartzell, Caroline and Matthew Hoddie. 2003. “Institutionalizing peace: Power sharing and post-civil war conflict management.” American Journal of Political Science 47(2): 318–332. Henderson, Sarah and Alana Jeydel. 2013. Women and Politics in a Global World. Oxford University Press. Henshaw, Alexis, June Eric-Udorie, Hannah Godefa, Kathryn Howley, Cat Jeon, Elise Sweezy, and Katheryn Zhao. 2019. “Understanding women at war: a mixed-methods exploration of leadership in non-state armed groups.” Small Wars and Insurgencies 30(6- 7): 1089–1116. Henshaw, Alexis Leanna. 2016. “Where Women Rebel: Patterns of women’s participation in armed rebel groups 1990 - 2008.” International Feminist Journal of Politics 18(1): 39–60. Henshaw, Alexis Leanna. 2020. “Female Combatants in Postconflict Processes: Understand- ing the Roots of Exclusion.” Journal of Global Security Studies 5(1): 63–79. Herrera, Natalia and Douglas Porch. 2008. “‘Like going to a fiesta’-the role of female fighters in colombia’s FARC-EP.” Small Wars and Insurgencies 19(4): 609–634. Hinojosa, Magda. 2012. Selecting Women, Electing Women: Political Representation and Candidate Selection in Latin America. Temple University Press. Huang, Reyko. 2016. The Wartime Origins of Democratization: Civil War, Rebel Gover- nance, and Political Regimes. Cambridge University Press. Huddy, Leonie and Nayda Terkildsen. 1993. “Gender Stereotypes and the Perception of Male and Female Candidates.” American Journal of Political Science 37(1): 119–147. Hughes, Melanie M. 2009. “Armed Conflict, International Linkages, and Women’s Parlia- mentary Representation in Developing Nations.” Social Problems 56(1): 174–204. Hughes, Melanie M. 2011. “Intersectionality, quotas, and minority women’s political repre- sentation worldwide.” American Political Science Review 105(3): 604–620. Hughes, Melanie M. and Aili Mari Tripp. 2015. “Civil War and Trajectories of Change in Women’s Political Representation in Africa, 1985-2010.” Social Forces 93(4): 1513–1540. Hughes, Melanie M. and Pamela Paxton. 2008. “Continuous Change, Episodes, and Critical Periods: A Framework for Understanding Women’s Political Representation over Time.” Politics and Gender 4(2): 233–264. Hughes, Melanie M., Pamela Paxton, Amanda B. Clayton, and Par Zetterberg. 2019. “Global gender quota adoption, implementation, and reform.” Comparative Politics 51(2): 219– 238. 143 Hultman, Lisa and Lisa Hultman. 2007. “Battle Losses and Rebel Violence : Raising the Costs for Fighting Battle Losses and Rebel Violence : Raising the Costs for Fighting.” 6553. IDEA. 2011. Women Members of the Constituent Assembly: A study on contribution of women in constitution making in Nepal. Technical report International Institute for Democracy and Electoral Assistance. Ishiyama, John. 2019. “Identity Change and Rebel Party Political Success.” Government and Opposition 54(3): 454–484. Ishiyama, John and Anna Batta. 2011a. “Rebel organizations and conflict management in post-conflict societies 1990-2009.” Civil Wars 13(4): 437–457. Ishiyama, John and Anna Batta. 2011b. “Rebel organizations and conflict management in post-conflict societies 1990-2009.” Civil Wars 13(4): 437–457. Ishiyama, John and Michael Marshall. 2015. “Candidate recruitment and former rebel par- ties.” Party Politics 21(4): 591–602. Ishiyama, John and Michael Marshall. 2017. “What explains former rebel party name changes after a civil conflict ends? External and internal factors and the transition to political competition.” Party Politics 23(4): 364–375. Ishiyama, John and Michael Widmeier. 2013. “Territorial Control, Levels of Violence, and the Electoral Performance of Former Rebel Political Parties After Civil Wars.”. Ishiyama, John and Michael Widmeier. 2020. “From “bush bureaucracies” to electoral com- petition: what explains the political success of rebel parties after civil wars?” Journal of Elections, Public Opinion and Parties 30(1): 42–63. Kalyvas, Stathis. 2006. The Logic of Violence in Civil War. Cambridge University Press. Kang, Alice J. and Aili Mari Tripp. 2018. “Coalitions Matter: Citizenship, Women, and Quota Adoption in Africa.” Perspectives on Politics 16(1): 73–91. K.C, Luna. 2019. “Everyday realities of reintegration: experiences of Maoist ‘verified’ women ex-combatants in the aftermath of war in Nepal.” Conflict, Security and Development 19(5): 453–474. K.C, Luna and Gemma Van Der Haar. 2019. “Living Maoist gender ideology: experiences of women ex-combatants in Nepal.” International Feminist Journal of Politics 21(3): 434– 453. Kittilson, Miki Caul. 2008. “Representing women: The adoption of family leave in compar- ative perspective.” Journal of Politics 70(2): 323–334. Kreutz, Joakim. 2010. “How and when armed conflicts end: Introducing the UCDP conflict termination dataset.” Journal of Peace Research 47(2): 243–250. 144 Krook, Mona Lena. 2010. “Beyond supply and demand: A feminist-institutionalist theory of candidate selection.” Political Research Quarterly 63(4): 707–720. Kunovich, Sheri and Pamela Paxton. 2005. “Pathways to power: The role of political parties in women’s national political representation.” American Journal of Sociology 111(2): 505– 552. Lahai, John. 2015. Gendering Conflict and Peacebuilding in Sierra Leone. In Female Com- batants in Conflict and Peace: Challenging Gender in Violence and Post-Conflict Reinte- gration. Springer. Loken, Meredith. 2018. Women in War: Militancy, Legitimacy, and Rebel Success PhD thesis. Lovenduski, Joni and Pippa Norris. 2003. “Westminster women: The politics of presence.” Political Studies 51(1): 84–102. Lu, Kelan and Marijke Breuning. 2014. “Gender and generosity: does women’s representation affect development cooperation?” Politics, Groups, and Identities 2(3): 313–330. Lührmann, Anna and Et. Al. 2020. “Varieties of Party Identity and Organization (V-Party) Dataset V1.”. Lyons, Tanya. 2004. Guns and Guerilla Girls: Women in the Zimbabwean National Libera- tion Struggle. Africa World Press. Lyons, Terrence. 2016. “From victorious rebels to strong authoritarian parties: prospects for post-war democratization.” Democratization 23(6): 1026–1041. MacKenzie, Megan. 2009. “Securitization and desecuritization: Female soldiers and the reconstruction of women in post-conflict Sierra Leone.” Security Studies 18(2): 241–261. Mama, Amina. 2020. “‘We will not be pacified’: From freedom fighters to feminists.” Euro- pean Journal of Women’s Studies 27(4): 362–380. Manekin, Devorah and Reed M. Wood. 2020. “Framing the Narrative: Female Fighters, External Audience Attitudes, and Transnational Support for Armed Rebellions.” Journal of Conflict Resolution pp. 1–28. Mann, Laura and Marie Berry. 2016. “Understanding the Political Motivations That Shape Rwanda’s Emergent Developmental State.” New Political Economy 21(1): 119–144. Manning, Carrie. 2004. “Armed Opposition Groups into Political Parties: Comparing Bosnia, Kosovo, and Mozambique.” Studies in Comparative International Development 39(1): 54–76. Manning, Carrie. 2007. “Party-Building on the Heels of War: El Salvador Bosnia, Kosovo and Mozambique.” Democratization 14(2): 253–272. 145 Manning, Carrie and Ian Smith. 2019. “Electoral Performance by Post-Rebel Parties.” Government and Opposition 54(3): 415–453. Marshall, Michael Christopher and John Ishiyama. 2016. “Does political inclusion of rebel parties promote peace after civil conflict?” Democratization 23(6): 1009–1025. Massaquoi, William N. 2007. “Women and Post-conflict Development: A Case Study on Liberia.” Gender 3(2): 134. Matanock, Aila M. 2016. “Using violence, seeking votes: Introducing the Militant Group Electoral Participation (MGEP) dataset.” Journal of Peace Research 53(6): 845–853. Matanock, Aila M. 2017. “Bullets for Ballots: Electoral Participation Provisions and En- during Peace After Civil Conflict.” International Security 41(4): 93–132. Matanock, Aila M. 2018. “External Engagement: Explaining the Spread of Electoral Partic- ipation Provisions in Civil Conflict Settlements.” International Studies Quarterly 62(3): 656–670. Mcdermott, Monika L. 1998. “Race and Gender Cues in Low-Information Elections.” Polit- ical Research Quarterly 51(4): 895–918. McKay, Susan. 2005. “Girls as “Weapons of Terror” in Northern Uganda and Sierra Leonean Rebel Fighting Forces.” Studies in Conflict and Terrorism 28(5): 385–397. Mechkova, Valeriya and Ruth Carlitz. 2020. “Gendered accountability: When and why do women’s policy priorities get implemented?” European Political Science Review (May). Meintjes, Sheila, Anu Pillay, and Meredith Turshen. 2002. There Is No Aftermath for Women. In The Aftermath: Women in Post-Conflict Transformation. pp. 3–18. Moran, Mary H. 2010. “Gender, Militarism, and Peace-Building: Projects of the Postconflict Moment.” Annual Review of Anthropology 39(1): 261–274. Mugabme, Beatrice. 2000. “Women’s Roles in Armed Conflict and Their Marginalisation in the Governance of Post-Conflict Society: The Case of Luwero Triangle.”. Muriaas, Ragnhild L. and Vibeke Wang. 2012. “Executive dominance and the politics of quota representation in Uganda.” Journal of Modern African Studies 50(2): 309–338. NDI. 2010. Marching Ahead: A Profile Book of Women CA Members. Technical report National Democratic Institute for International Affairs. Ni Aolain, F. and E. Rooney. 2007. “Underenforcement and Intersectionality: Gendered Aspects of Transition for Women.” International Journal of Transitional Justice 1(3): 338–354. Nilsson, Johanna and Suruchi Thapar-Björkert. 2013. “People constantly remind me of my past. and make me look like a monster.” International Feminist Journal of Politics 15(1): 110–118. 146 Niven, David. 1998. “Party elites and women candidates: The shape of bias.” Women and Politics 19(2): 57–80. Norris, Pippa and Joni Lovenduski. 1993. “‘If Only More Candidates Came Forward’: Supply-Side Explanations of Candidate Selection in Britain.” British Journal of Politi- cal Science 23(3): 373–408. Norris, Pippa and Joni Lovenduski. 1995. Political Recruitment: Gender, Race and Class in the British Parliament. Cambridge: Cambridge University Press. O’Brien, Diana and Joanna Rickne. 2014. “Gender Quotas and Women’s Political Leader- ship.” Research Institute of Industrial Economics (IFN) IFN Workin. Pankhurst, Donna. 2003. “The ’sex war’ and other wars: Towards a feminist approach to peace building.” Development in Practice 13(2-3): 154–177. Pankhurst, Donna. 2012. Gendered Peace: Women’s Struggles for Post-War Justice and Reconciliation. Routledge. Paxton, Pamela, Melanie M. Hughes, and Matthew A. Painter. 2010. “Growth in women’s political representation: A longitudinal exploration of democracy, electoral system and gender quotas.” European Journal of Political Research 49(1): 25–52. Paxton, Pamela and Sheri Kunovich. 2003. “Women’s political representation: The impor- tance of ideology.” Social Forces 82(1): 87–114. Paxton, Pamela, Sheri Kunovich, and Melanie M. Hughes. 2007a. “Gender in politics.” Annual Review of Sociology 33: 263–284. Paxton, Pamela, Sheri Kunovich, and Melanie M. Hughes. 2007b. “Gender in Politics.” Annual Review of Sociology 33(1): 263–284. Paxton, Pamela, Sheri Kunovich, and Melanie M. Hughes. 2007c. “Gender in Politics.” Annual Review of Sociology 33(1): 263–284. Pet, Nils, T E R Gleditsch, Mikael Eriksson, Håvard Strand, Meredith Reed Sarkees, and Dan Smith. 2002. “Armed Conflict 1946 – 2001 : A New Dataset.” Journal of Peace Research 39(5): 615–637. Pettersson, Therése and Magnus Öberg. 2020. “Organized violence, 1989–2019.” Journal of Peace Research 57(4): 597–613. Phillips, Anne. 1995. The Politics of Presence. Clarendon Press. Rosen, Jennifer. 2013. “The Effects of Political Institutions on Women’s Political Represen- tation: A Comparative Analysis of 168 Countries from 1992 to 2010.” Political Research Quarterly 66(2): 306–321. Salehyan, Idean, Kristian Skrede Gleditsch, and David E. Cunningham. 2011. “Explaining External Support for Insurgent Groups.” International Organization 65(4): 709–744. 147 Sanbonmatsu, Kira. 2002. “Political Parties and the Recruitment of Women to State Legis- latures.” The Journal of Politics 64(3): 791–809. Sanı́n, Francisco Gutiérrez and Francy Carranza Franco. 2017. “Organizing women for combat: The experience of the FARC in the Colombian war.” Journal of Agrarian Change 17(4): 770–778. Schwindt-Bayer, Leslie A. and S. Alles. 2018. Women in legislatures: Gender, institutions, and democracy. Oxford University Press. Seidman, Gay W. 1984. “Women in Zimbabwe: Post Independence Struggles.” Feminist Studies 10(3): 419–440. Shair-Rosenfield, Sarah and Reed M. Wood. 2017. “Governing well after war: How improving female representation prolongs post-conflict peace.” Journal of Politics 79(3): 995–1009. Shekhawat, Seema. 2015. Visible in Conflict, Invisible in Peace: Positioning Women in the Militancy in Kashmir. In Female Combatants in Conflict and Peace: Challenging Gender in Violence and Post-Conflict Reintegration. Sindre, Gyda M. 2019. “Adapting to Peacetime Politics? Rebranding and Ideological Change in Former Rebel Parties.” Government and Opposition 54(3): 485–512. Sindre, Gyda Marås. 2016a. “In Whose Interests? Former Rebel Parties and Ex-Combatant Interest Group Mobilisation in Aceh and East Timor.” Civil Wars 18(2): 192–213. Sindre, Gyda Marås. 2016b. “Internal party democracy in former rebel parties.” Party Politics 22(4): 501–511. Sjoberg, Laura. 2010. “Women fighters and the ‘beautiful soul’ narrative.” International Review of the Red Cross 92(877): 53–68. Sjoberg, Laura. 2018. “Jihadi brides and female volunteers: Reading the Islamic State’s war to see gender and agency in conflict dynamics.” Conflict Management and Peace Science 35(3): 296–311. Sjoberg, Laura and Carol Gentry. 2007. Mothers, Monsters, Whores: Women’s Violence in Global Politics. Zed Books. Söderberg Kovacs, Mimmi. 2008. When rebels change their stripes: Armed insurgents in post-war politics. In From War to Democracy: Dilemmas of Peacebuilding. Vol. 9780521885 pp. 134–156. Söderberg Kovacs, Mimmi. 2019. “The Legacy of a Revolution that Never Happened: The Post-War Politics of Former Rebel Party RUFP in Sierra Leone.” Government and Oppo- sition pp. 1–15. Söderberg Kovacs, Mimmi and Sophia Hatz. 2016. “Rebel-to-party transformations in civil war peace processes 1975–2011.” Democratization 23(6): 990–1008. 148 Söderström, Johanna. 2013. “Second time around: Ex-combatants at the polls in Liberia.” Journal of Modern African Studies 51(3): 409–433. Tamale, Sylvia. 1999. When Hens Begin to Crow: Gender and Parliamentary Politics in Uganda. Boulder, CO: Westview Press. Thomas, Jakana L. 2020. “Wolves in Sheep’s Clothing: Assessing the Effect of Gender Norms on the Lethality of Female Suicide Terrorism.” International Organization . Thomas, Jakana L. and Kanisha D. Bond. 2015. “Women’s Participation in Violent Political Organizations.” American Political Science Review 109(3): 488–506. Thomas, Jakana L. and Reed M. Wood. 2018. “The social origins of female combatants.” Conflict Management and Peace Science 35(3): 215–232. Tickner, Ann J. 1999. “Why women can’t run the world: International politics according to Francis Fukuyama.” International Studies Review 1(3): 3–11. Tripp, A. M. 2001. “The politics of autonomy and cooptation in Africa: The case of the Ugandan Women’s movement.” Journal of Modern African Studies 39(1): 101–128. Tripp, Aili Mari. 2016. “Women’s mobilisation for legislative political representation in Africa.” Review of African Political Economy 43(149): 382–399. Tripp, Aili Marl. 2000. Women and Politics in Uganda. University of Wisconsin Press. Tripp, Aili Marl. 2015. Women and Power in Post-Conflict Africa. Cambridge University Press. Tull, Denis M. and Andreas Mehler. 2005. “The hidden costs of power-sharing: Reproducing insurgent violence in Africa.” African Affairs 104(416): 375–398. Turshen, Meredith. 2002. Engendering Relations of State to Society in the Aftermath. In The Aftermath: Women in Post-Conflict Transformation. pp. 78–98. Utas, Mats. 2003. “Sweet Battlefields: Youth and the Liberian Civil War.”. Valdini, Melody E. 2019. The Inclusion Calculation: Why Men Appropriate Women’s Rep- resentation. Oxford University Press. Valdini, Melody Ellis. 2013. “Electoral institutions and the manifestation of bias: The effect of the personal vote on the representation of women.” Politics and Gender 9(1): 76–92. van de Goor, Luc and Jeroen de Zeeuw. 2008. “International Involvement in Rebel-to-Party Transformations.” From Soldiers to Politicians: Transforming Rebel Movements After Civil War pp. 225–254. Viterna, Jocelyn. 2013. Women in War: The Micro-Processes of Mobilization in El Salvador. Oxford University Press. 149 Viterna, Jocelyn. 2014. Radical or Righteous? Using Gender to Shape Public Perceptions of Political Violence. In Dynamics of Political Violence : A Process-Oriented Perspective on Radicalization and the Escalation of Political Conflict. Viterna, Jocelyn, Kathleen M. Fallon, and Jason Beckfield. 2008. “How development mat- ters: A research note on the relationship between development, democracy and women’s political representation.” International Journal of Comparative Sociology 49(6): 455–477. Wang, Vibeke and Ragnhild Louise Muriaas. 2019. “Candidate selection and informal soft quotas for women: gender imbalance in political recruitment in Zambia.” Politics, Groups, and Identities 7(2): 401–411. Waylen, G. 2004. “What can the South African transition tell us about gender and democ- ratization.” Unpublished manuscript, Queens University Belfast (September): 1–21. Weinstein, Jeremy. 2006. Inside Rebellion: The Politics of Insurgent Violence. Cambridge University Press. Wittig, Katrin. 2016. “Politics in the Shadow of the Gun: Revisiting the Literature on ‘Rebel-to-Party Transformations’ through the Case of Burundi.” Civil Wars 18(2): 137– 159. Wood, Reed. 2019. Female Fighters: Why Rebel Groups Recruit Women for War. Columbia University Press. Wood, Reed M. 2014. “Opportunities to kill or incentives for restraint? Rebel capabilities, the origins of support, and civilian victimization in civil war.” Conflict Management and Peace Science 31(5): 461–480. Wood, Reed M. and Jakana L. Thomas. 2017. “Women on the frontline: Rebel group ideology and women’s participation in violent rebellion.” Journal of Peace Research 54(1): 31–46. Wooldridge, Jeffrey M. 2010. Econometric Analysis of Cross-Section and Panel Data. 2nd ed. MIT Press. Wooldridge, Jeffrey M. 2012. Introductory Econometrics: A Modern Approach. Fifth ed. Cengage Learning. Wylie, Kristin. 2018. Party Institutionalization and Women’s Representation in Democratic Brazil. Cambridge University Press. Yadav, P. 2020. “Can women benefit from war? Women’s agency in conflict and post-conflict societies.” Journal of Peace Research pp. 1–13. Yadav, Punam. 2016. Social Transformation in Post-Conflict Nepal: A Gender Perspective. New York: Routledge. Yadav, Punam. 2017. Women in the Parliament: Changing Gender Dynamics in the Political Sphere in Nepal. In Women in Government Institutions in South Asia. pp. 31–53. 150 Yoon, Mi Yung. 2004. “Explaining Women’s Legislative Representation in Sub-Saharan Africa.” Legislative Studies Quarterly 29(3): 447–468. Zaks, Sherry. 2017. Resilience Beyond Rebellion: How Wartime Organizational Structures Affect Rebel-to-Party Transformation PhD thesis. 151