Doctoral Dissertation Research in Political Science: Rebel Organization and Civilian Abuse

在过去半个世纪的内战中，约有1800万人丧生。​战争的这些间接代价最终比战斗本身更具破坏性。但是，平民和士兵支付的相对费用有很大差异。例如，在斯里兰卡，平民与士兵的死亡比例为1:1；而在刚果民主共和国，这一比例超过了100:1。这就引出了一个问题，为什么一些反叛组织在他们的领土上建立秩序，而另一些则虐待和掠夺平民。政治科学对这个问题的最新答案主要集中在叛军如何获得资金以及叛军与国家民兵之间的战略平衡上。本文从反叛组织的内部动力学出发，提出了一种新的反叛群体行为理论。它假设叛军领导人有经济动机在他们的领土上创造秩序，但往往无法阻止组织成员抢劫和虐待平民。虽然早期的理论认为资源与掠夺行为之间存在正相关关系，但该项目的理论预测，当反叛领导人获得财务资源时，他们将对下属施加更有效的控制。第二个预测抑制掠夺行为的因素是领导人和他们的指挥官之间密切的种族/社会关系。最后，该项目确定了迄今为止在文献中被忽视的第三个因素。研究人员发现，外部赞助人的目标很重要，在掠夺性行为的情况下，外国供应商有动机支持更容易被操纵的弱势领导人，从而导致滥用、派系化的团体。该模型应用于1989年至2003年的利比里亚内战，根据上述因素预测了利比里亚反叛组织之间的变化。为了测试该模型，该项目使用了地球科学开发的遥感方法。通过处理卫星图像来衡量利比里亚战争期间农作物面积的变化，以此作为平民虐待的代表。这些作物估计是结合了14个月的实地调查收集到的有关反叛组织特征的系统数据，以及地理和人口数据。初步结果表明，随着时间的推移，作物面积发生了显著变化，这与模型的预测相一致。该模型在1945-2003年反叛组织的原始跨国数据集上进行了进一步测试。由于缺乏可靠的微观数据，对内战的研究一直受到限制。本文开发了一种严格的方法来克服这些限制，同时也为该领域提供了理论贡献。通过强调粮食安全和集团内谈判，本文为设计人道主义干预和成功的和平协议提供了具体的含义。

In civil wars over the past half-century, some 18 million people lost their lives. In many of these wars, despite a Mao Tse-tung image of rebels winning hearts and minds, civilians pay the heaviest cost, with high levels of displacement, malnutrition, and disease. These indirect costs of war are ultimately more damaging than the fighting. But there is significant variation in the relative costs paid by civilians and those paid by soldiers. For example, in Sri Lanka, the ratio of civilian to soldier deaths was 1:1; meanwhile in the Democratic Republic of Congo, that ratio was more than 100:1. This leads to the question of why some rebel groups create order in their territories, while others abuse and prey upon the civilian population. State-of-the-art answers to this question in political science focus mostly on how rebels get funded and on the strategic balance between rebels and state militias.The dissertation proposes a new theory of rebel group behavior based on the internal dynamics of rebel organizations. It assumes that rebel leaders have an economic incentive to create order in their territories but are often unable to prevent group members from looting and abusing the civilian population. While earlier theories saw a positive relationship of resources and predatory behavior, this project's theory predicts that rebel leaders will exert more effective control over their subordinates when they have access to financial resources. A second factor predicting restraint on predatory behavior is that of close ethnic/social ties between leaders and their commanders. Finally, the project has identified a third factor so far ignored in the literature. The researcher finds that the goals of external patrons matter, and in cases of predatory behavior, there are incentives by foreign suppliers to support weak leaders who can be more easily manipulated, resulting in abusive, factionalized groups.Applied to the civil war in Liberia, 1989-2003, this model predicts variation among Liberia's rebel groups, depending on the factors listed above. To test the model, the project uses a remote sensing methodology developed in the earth sciences. Satellite images are processed to measure changes in crop area as a proxy for civilian abuse during Liberia's war. These crop estimates are combined with systematic data on rebel group characteristics collected through 14 months of field work, as well as geographic and demographic data. Preliminary results indicate significant variation in crop area over time, consistent with the model's predictions. The model is further tested on an original, cross-national dataset of rebel groups, 1945-2003.The study of civil war has been constrained by a lack of reliable, micro-level data. This dissertation develops a rigorous method for overcoming these limitations while also offering a theoretical contribution to the field. Through its emphasis on food security and intra-group bargaining, the dissertation provides concrete implications for designing humanitarian interventions and successful peace agreements.