Research

Mediation is a crucial instrument employed by external actors to resolve armed conflicts and mitigate violence. A large academic literature examines mediation in civil war, with analyses of which civil wars see mediation and what the effect of this mediation is. Many organizations express a commitment to conflict prevention, and engage in mediation to prevent the outbreak of armed conflict. There is much less research on when mediation is used as a tool of conflict prevention. We have collected new data on all mediation efforts in a random sample of 51 self-determination (SD) disputes from 1991–2015, which includes disputes that never experience armed conflict, as well as years before and after armed conflicts in disputes that do. We use these data to examine the conditions under which preventive mediation occurs in SD disputes. We develop a theoretical argument for when mediators are likely to offer mediation, and governments and representatives of SD groups are likely to accept it. We test this argument using our new data in two samples of SD dispute-years that are not in armed conflict. We find that mediation is more likely in dispute-years outside of armed conflict where SD groups are engaged in low-level violence and in disputes in countries that border other countries that are experiencing armed conflict and less likely in disputes in states that are permanent members of the UN Security Council or former French colonies. This analysis shows that mediators do engage in preventive mediation in disputes that they perceive as having a higher likelihood of escalation to armed conflict, but that they are constrained in their ability to do so by geopolitics.

This paper introduces new data on the creation of subsidiary bodies (SBs) by members of the United Nations Security Council (UNSC) between 1972 and 2020. Delegation to SBs is one of the principal means through which the UNSC acts, and these bodies are designed to carry out crucial functions such as peacekeeping, implementing sanctions, and investigating crises. Yet, no research has systematically evaluated their creation, design, and use. Our dataset includes a typology of all proposed and created SBs as well as information about their purpose and design. After introducing the data, we empirically analyze the determinants of SB creation. Multivariate regression demonstrates that SBs are more likely to be created when the preferences of the permanent members are aligned. Moreover, stronger bodies are more likely to be created during periods of high preference alignment, while middle- and lower-strength bodies are less influenced by member alignment. These results provide unique evidence demonstrating how politics affects the choice of when and how the UNSC responds to global problems. Our data and analysis paint a picture of a more proactive UNSC than is commonly portrayed in the literature, and these data will enable scholars to further analyze UNSC action.

The United Nations Security Council responds to mass atrocity events with striking inconsistency, despite its mandate to maintain international peace and security. Some crises receive resolutions authorizing robust responses, including peacekeeping operations with civilian protection mandates. In others, resolutions are proposed but fail to pass, or only offer rhetorical rebuffs. In still others, most starkly illustrated by the end of the Sri Lankan Civil War, the UNSC never meets at all. Understanding this inconsistency requires disaggregating what existing research typically treats as a single process: the decisions to propose and pass a resolution. I argue that these stages reflect different incentive structures. Proposing a resolution is associated with reputational concerns related to political distance from the target state, distribution of preferences among Council members, and international attention. Alternatively, passage requires collective agreement constrained by great-power politics. Using a dataset of 394 UNSC meetings on mass atrocity events from 1989 to 2024, analyzed at the state and meeting level, I show that political distance and preference variance are associated with proposals, while passage is associated with great-power preference alignment. I find that states most distant from the target state are most likely to draft a resolution precisely when the Council is most divided, consistent with the interpretation that proposal clarifies their position to international audiences and forces co-members to publicly reveal theirs. Failed resolutions thus function as expressive political acts rather than institutional failure, helping explain why the Council can appear simultaneously active yet ineffective.

Why does the UN Security Council frequently convene on some mass atrocities while others are left off the docket entirely? Since October 2023, Gaza has occupied a near-constant place on the Council's formal agenda, despite deep and persistent division among members. Fifteen years earlier, a comparably severe endgame to Sri Lanka's civil war never reached the agenda at all. Existing accounts of UNSC agenda-setting, built around the Council's ordinary docket, expect preference divergence to suppress engagement: where members disagree, resolutions are unlikely, and issues that cannot produce one are not worth the Council's limited plenary time. This logic sits uneasily with cases like Gaza, where frequent engagement, as well as frequent vetoes, is present. This paper argues that mass atrocity (MA) events call for a different logic, for two reasons. First, MA events clear a normative severity threshold by construction, making non-response a visible and reputationally costly failure rather than a neutral scheduling choice. Second, drawing on the Council's function as a collective legitimizer, I relax the assumption that resolution passage is the meaningful marker of engagement. A formal meeting is itself a consequential act of collective judgment, independent of its legislative output. Once passage is set aside as the object members are presumed to pursue, preference variance may no longer act as a brake on agenda-setting. I argue instead that, for MA events, variance operates as a gas pedal: division raises the reputational stakes of silence and gives Council members more reason to convene, and to convene repeatedly, even without consensus. I test these expectations using an original dataset of UNSC responses to 24 mass atrocity events with violence onset between 1989 and 2022.