Research

Publications

1. Boyes, Christina, Tiina Hyyppä, Merve Keskin, Dana M. Landau, Rachel A. Schwartz, Megan A. Stewart, Alex Waterman, and James Worrall. 2025. “Civil Wars ISA Roundtable 2025 Civil War and Intrastate Conflict: Reimagining a Field of Study.” Civil Wars, 27(2), 416–439.
2. Zhang, Xiaozhong, Jiawei Xu, Merve Keskin, Michael Colaresi, Vladimir Zadorozhny, and Panos Chrysanthis. 2022. “HR-VEAW: A Human Rights Violation Exploration, Analytics, and Warning System.” International Workshop on Big Data Visual Exploration and Analytics.
   Read more

   The availability of information from social media, such as tweets, and human rights monitors like Amnesty International, Human Rights Watch, and the U.S. State Department has created new opportunities to measure repression and human rights protections in higher resolution. In this paper, we present HR-VEAW, a Human Rights Violation Exploration, Analytics, and Warning System, designed to support the understanding of social conflict dynamics and human rights violations or protections using quantitative data. After briefly discussing HR-VEAW’s data acquisition and analysis components, we demonstrate how it visualizes rich spatio-temporal and conceptual information, enabling researchers to examine changes in patterns of violation and protection over time or across both space and time. HR-VEAW helps explain social instability and conflict, and guides decision-making, theorizing, and predictive analysis.

Works in Progress

1. Forecasting Local Agreements in Civil War: A Machine Learning Perspective (Under Review)
   Read more

   “While substantial scholarship examines national-level peace settlements in civil wars, the role of local peace agreements in shaping rebel-civilian interactions remains understudied. Given that we lack research identifying the functional form linking the balance of legitimacy and security at the local level, I employ flexible machine learning algorithms alongside time- series cross-validation that do not impose specific linear forms on the response functions. My theoretical framework naturally lends itself to a predictive approach. Using a novel dataset from the Central African Republic civil war, my approach conditions forecasts on varying feature sets. My results indicate that territorial contestation, distance to the capital, and proximity to economic resources strongly predict the formation of local peace agreements, suggesting that these agreements arise from deliberate calculations by armed groups and civilians. In addition, my analysis reveals that the regularized (log) logistic model performs more weakly than flexible machine learning models, indicating the presence of nonlinearities and contingent relationships among features. Ultimately, these findings underscore that local agreements are adaptive strategies that shape rebel-civilian dynamics in conflict zones, emphasizing the importance of subnational and predictive approaches in conflict management."

2. Between Peacebuilding and Peacemaking: China’s Evolving Role in Conflict Management (with Monalisa Adhikari) (Under Review)
   Read more

   “China’s growing involvement in global peace and security has generated debate over its distinctive approach to peacebuilding and its expanding role as a peacemaker. Drawing on an original survey of 3,128 citizens and 47 elite interviews in Myanmar, this article examines how citizens perceive China’s peacebuilding and peacemaking roles and how these views shape its perceived effectiveness as a mediator. The analysis shows that key features of China’s peace building approach, its emphasis on development as a pathway to peace and its state-centric mode of engagement, strongly influence perceptions of China’s impartiality and leverage. Framing infrastructure and investment projects as developmental peace has led many citizens to view China as a vested actor focused on protecting its economic interests rather than resolving conflict, thereby undermining perceptions of impartiality. Moreover, as such investments create uneven benefits, they diminish China’s leverage over all conflict parties. China’s reliance on state institutions further reinforces perceptions of partiality, with many seeing it as aligned with the government. By foregrounding citizen perspectives, the article shows how China’s peacebuilding practices constrain its credibility and influence in peacemaking, challenging the assumption that peacebuilding and peacemaking are necessarily mutually reinforcing."

3. International Rivalry, Military Intervention, and Duration of Civil War (with Jungmin Han) (Under Review)
   Read more

   “Civil wars often unfold amid complex networks of foreign involvement. Although prior research shows that two-sided interventions, where both governments and rebel groups receive external support, tend to prolong conflicts, it typically treats them as interactions between two unified camps, overlooking the distinct roles and linkages of individual intervening states. We argue that the structure of external competition, specifically, how intervening states are networked across opposing camps, fundamentally shapes conflict duration. The analysis introduces the concept of a competitive intervener dyads (CID), defined as pairs of states supporting opposite sides in a civil war. As the number of these dyads increases, the density of cross-camp competition rises, complicating bargaining and reducing the likelihood of settlement. Moreover, when CIDs consist of interstate rivals, mutual mistrust and antagonism further undermine credible commitments and intensify informational asymmetries, making peace even less attainable. Using global data on civil wars from 1975 to 2017, combined with measures of external support and interstate rivalry, we estimate Cox proportional hazards models to test these propositions. The results show that a higher number of CIDs significantly decreases the tendency of civil war to terminate, and that rivalry dyads exert particularly strong prolonging effects. These findings highlight how the configuration of external actors, not merely their presence, profoundly shapes the trajectory of civil wars, bridging research on foreign intervention, bargaining dynamics, and international rivalry."

4. A Rebel with a Sanction: UN Sanctions against Non-state Armed Groups and Civil Conflict Duration (with Omer Zarpli) (Under Review)
   Read more

   “How do economic sanctions against non-state armed groups affect civil conflict duration? Since the end of the Cold War, there has been an increase in the use of economic sanctions against violent non-state actors. While a limited number of studies have explored the links between economic sanctions and civil wars, they have focused exclusively on sanctions imposed on governments. We have limited systematic evidence on whether sanctions targeting non-state actors effectively contribute to resolving armed conflicts. We argue that sanctions against rebel groups prolong conflict duration. The enforcement problems that plague sanctions exacerbate uncertainties about future relative power between the combatants. This, in turn, creates incentives to continue to fight in order to accrue more information. However, the conflict-prolonging effect of sanctions is immediate and dissipates over time as combatants adapt to the new conditions imposed by sanctions. We test these hypotheses on a sample of United Nations-authorized sanctions on rebel groups between 1990 and 2024 and find strong empirical support. The results are robust to alternative model specifications. Supplementary analyses aimed at testing the theoretical mechanism further support our findings."

5. Measuring Child Recruitment in Civil Wars Using Large Language Models: A New Actor-Level Dataset (with Isil Idrisoglu and Rebecca Cordell)
   Read more

   “Why do some governments and rebel groups recruit children while others do not? Which actor, country, and conflict characteristics shape how and where children are used, the types of children recruited, and the intensity of forced recruitment in civil wars? This puzzle has been widely studied in conflict research, yet existing data remain limited—typically binary, static, and lacking detail on actors, timing, or roles. To fill this gap, we construct a new actor-level dataset on child involvement in conflict and international law using a range of large language models (LLMs) to extract and code information from UN documents. We compile 400 reports from 21 conflict-affected countries (1996–2025), convert them to machine- readable text, segment over 60,000 paragraphs, and hand-code a sample. This training data is used to 1) fine-tune an LLM for classification; 2) validate named-entity recognition and dictionary-based models; 3) evaluate generative AI models; and 4) test Retrieval-Augmented Generation. We then apply the best-performing model to the full corpus. The resulting dataset captures perpetrators, recruitment tactics, children’s roles, legal violation types, location, age, gender, and scale. This new data resource will enable analysis of the link between the demand and supply side of recruitment dynamics and legitimacy concerns, offering a foundation for future research on child soldiering and human rights in conflict."

6. Child Recruitment and Rebel Groups' Legitimacy Seeking in Civil Wars (with Isil Idrisoglu and Rebecca Cordell)
7. Linking Forecasting and Conflict Data: Peace Agreements and Violence Dynamics (with Michael P. Colaresi, Qing Chang, Laura Chelidonopoulos, Zhejun Qui)
8. Who Should Be Included in Peace Negotiations? Public Opinion from Myanmar (with Monalisa Adhikari)
9. Peace or Retribution? The Role of Warfare on Public Opinion Toward Peace in Colombia
10. Mitigating Violence Locally: The Effects of Local Agreements on Violence in Civil Wars