Attempts by elites to negotiate an end to civil conflict are often met with backlash and polarization in the public sphere. A large literature has examined why the choice to negotiate itself may incite a negative response from the public, yet less work has tried to causally disentangle what aspects of a settlement render it unpopular. I present a theory of public support for negotiated settlements that links individual predispositions to settlement terms bearing on retributive justice. Using novel conjoint experiments fielded during the Colombian peace process, I find evidence that agreements low on retributive justice are broadly unpopular, though negative reactions are largely dependent on peoples' fundamental world-views. The results have implications for the dynamics of polarization in societies at war.

One of the theorized pathways of Global Performance Assessment (GPA) influence is transnational pressure applied to poorly performing states.  This article provides empirical evidence of this mechanism at work.  We show that when a state receives a negative assessment in Freedom House's annual Freedom in the World report, it is treated systematically worse by other states in a phenomenon we term the ``Scarlet Letter effect.''  A state branded ``Not Free'' by Freedom House receives increased negative verbal interaction from norm-respecting democracies, especially in the months immediately following the annual release of the Freedom in the World report.  We separate the effect of Freedom House's assessment itself from the effect of the underlying factors Freedom House measures by exploiting a discontinuity in the assignment of a country's freedom status, whereby countries on either side of a bright-line receive different statuses, even though their level of political and civil freedom is extremely similar.  Additionally, our quantitative tests are supported by evidence gathered during interviews with senior officials at Freedom House as well as organizations that make use Freedom House's scores when making or advising policy.   We find that Freedom House's annual assessment matters -- not only because it informs governments and activists around the globe, but because it effects one of the most fundamental aspect of international relations: how states talk about each other. 

Life on the front-lines of a civil war is markedly different from life in safe(r) areas. How does this drastic difference in lived experience shape civilian attitudes towards war and peace? Contrary to theories that link conflict exposure to intransigence in war-torn societies, I argue that under certain conditions exposure increases support for both peace as an outcome as well as for the granting of concessions to armed actors that render settlement more likely. I use various model specifications and matching methodology on survey data from the Colombian peace process, finding evidence that civilians in conflict-zones exhibit greater support for the peace process overall and are more willing to grant political concessions to armed groups, yet less willing to personally reintegrate with demobilized fighters. The study has theoretical implications for accounts of conflict exposure and helps explain regional variation in the failed referendum vote in Colombia.  

  • “Elections and Violence in Civil Wars: Evidence from Colombia”. 

In spite of the dramatic toll political conflicts take on civil society, elections are often held in countries experiencing ongoing conflicts. Yet, there is still much we do not know about how elections and the electoral process interact with ongoing armed conflicts, particularly with regard to how armed actors interact with the civilian population. How do elections shape the course of violence in civil wars, and what is the electoral logic of that violence? This paper proposes that links to constituent groups and other interests incentivize armed groups to respond violently to the electoral process, and further that local electoral dynamics shape when and how armed actors use violence against the civilian population. I argue close electoral contests promote the targeting of actors with high electoral value (such as activists, social movements, or political actors) whose removal can decisively determine elections. Implications of the theory are tested in the context of the Colombian conflict using novel, fine-grained data on the targeting of labor unions. Results suggest that electoral dynamics clearly matter in shaping the use of force in civil wars, though the timing of violence is often most intense between elections than within them. 

Conflict negotiations often spark polarization in the public sphere. An extensive literature has uncovered a broad range of determinants for why civilians support or oppose peace processes, yet little work seeks to evaluate how well these determinants actually predict conflict-termination preferences or how important these determinants are relative to one another. We rely on machine learning methods and a large set of nationally-representative surveys from Colombia for the years 2004-2015 to explore which attributes are the strongest predictors of support for negotiating an end to war with the group's largest armed group, the FARC. We find that living in proximity to conflict, individual values related to justice and punishment, and elite influence are among the strongest predictors of negotiation preferences. The analysis sheds light on which factors drive wartime public opinion and speaks to the recent referendum results in the country.

Does the networked structure of security ties in the international system inform the extent of conflict and cooperation among states? This study sets out to do two things. First, it defines a set of joint-production security communities (JPSCs) in the international system and measures the extent of hierarchy within those communities. Second, it uses these community definitions and the measures of hierarchy to help explain patterns of conflict and cooperation within the communities and between them. Without a claim to the causal direction, we conclude that JPSCs are associated with more peaceful interstate relations, especially when the joint security production within the communities are more hierarchical. 

  •  “How to Fix IR: A Network Approach”, with  S. Minhas, C. Dorff, M. Foster, M. Gallop, H. Liu, and M. Ward.

International relations scholarship is filled with dyads. This includes dyadic hypotheses and propositions, but especially data. These relational data contain information about the interdependencies of various phenomena, including countries, dyads, and even triads. However, most empirical studies of such data do not take into account these dependencies. As a result, the studies are often contradictory and produce results that are not compelling. One reason is that the independence required by the typical methods employed does not exist in the data being studied. We present a different, regression based method, which constructs a latent network which incorporates first, second, and third-order dependencies. We replicate four prominent studies in recent IR scholarship and compare the standard approach to the latent factor approach. The additive and multiplicative latent factor approach is shown to produce more precise estimates of covariate effects, and it also dominates standard approaches in terms of out-of-sample cross-validations.