Research Agenda and Publications

This working paper replicates applied studies that use process tracing to assess how much their conclusions depend on qualitative interpretations of evidence. The design combines Bayesian process tracing with Van Evera’s test typology (hoop, smoking gun, doubly decisive, and straw-in-the-wind), translating alternative evidence readings into explicit probabilistic updates. The goal is to estimate how sensitive published inferences are to plausible choices about coding and evidentiary strength. The project aims to make qualitative causal diagnosis more transparent, comparable, and reproducible.

This working paper adapts posterior predictive checks to Bayesian process tracing to evaluate whether the likelihood specification used to model evidence is empirically plausible. The framework compares observed evidence patterns with model-generated predictive distributions to detect systematic mismatches between data and specification. When lack of fit appears, the approach provides diagnostics to revise parameters and likelihood structure. The central contribution is a formal validation routine that strengthens calibration, transparency, and confidence in qualitative causal inference.

How do trade-based status gains affect foreign policy alignment? Existing IR research emphasizes status anxiety and status seeking, but pays little attention to upward status shifts. We argue that rank reversals in trade hierarchies, especially when a rising power displaces the hegemon, create a salient status shock that can produce foreign policy realignment beyond gradual trade growth. We test this claim in Brazil, where China overtook the United States as top trade partner in 2009, using Synthetic Difference-in-Differences on UNGA ideal-point distance. The estimate implies a statistically significant 41% reduction (approximately 0.10 standard deviations) in Brazil’s ideological distance to China after the reversal (p = 0.032). To probe mechanism implications, we analyze Brazil’s main newspaper with NLP and find evidence consistent with a post-2009 increase in China-related salience, particularly trade-focused coverage. We then examine cross-country cases in which China displaced the United States as top partner. Because four of the eight treated countries experience treatment reversals (the USA regains the top position), we employ the counterfactual estimator of Liu, Wang, and Xu (2024) with interactive fixed effects, which accommodates non-absorbing (switching) treatment. The cross-country estimate confirms a similar pattern. Exit-aligned analysis further shows that the alignment effect dissipates when the USA regains the top position, providing evidence consistent with a salience-driven mechanism. These results suggest that discrete status milestones, not only continuous trade deepening, can reshape foreign policy behavior. The findings challenge claims that status gains are mostly illusory and motivate research on status effects in trade, investment, aid, and other economic ties.

The selection of control variables is a critical step in observational studies, yet standard heuristics can inadvertently lead researchers to include collider variables, introducing bias rather than removing it. While Directed Acyclic Graphs (DAGs) can diagnose whether a variable is a collider, they do not quantify the magnitude of the resulting bias. We derive the Included Variable Bias (IVB) formula—a closed-form expression that mirrors the classic Omitted Variable Bias (OVB) formula but addresses the opposite problem: the bias from erroneously including a variable. We show that IVB = −θ⋆ × π, where both components are directly estimable from data. We extend the result from cross-sectional models to the Autoregressive Distributed Lag (ADL) specifications common in time-series cross-sectional (TSCS) analysis. Monte Carlo simulations confirm the formula’s exactness across multiple data-generating processes, and we illustrate its application with an example from the civil war literature.

This paper examines the effect of bottom-up accountability on public service delivery. We differentiate between information-driven interventions and the mobilization and monitoring efforts of organized Civil Society Organizations (CSOs), and argue that the latter type of interventions can drive significant policy change. The study evaluates the effectiveness of the Obra Transparente project by Brazilian NGO Transparência Brasil, engaging 21 local CSOs in South and Southeast Brazil. Using a difference-in-differences design, we estimate that the intervention increased construction completion rates by approximately 8 percentage points on average (static specification), with suggestive evidence that the dynamic effect may reach up to 18 percentage points approximately five years after the start of the intervention, though this estimate should be interpreted with caution due to a gap in data collection spanning the COVID-19 pandemic. Our findings emphasize the importance of sustained, coordinated efforts by socially embedded CSOs. These efforts, involving direct engagement with municipal officials and ensuring that their complaints cannot be ignored, lead to more substantial outcomes when compared with information-driven interventions targeting individual citizens.

Este estudo investiga a influência dos fundamentos morais no engajamento político em mídias sociais em contextos não eleitorais no Brasil. Utilizando a Teoria dos Fundamentos Morais (MFT), exploramos como apelos morais direcionados — como autoridade e lealdade para públicos conservadores — impactam os níveis de engajamento (curtidas, comentários, compartilhamentos) em plataformas como Facebook e Instagram. Em parceria com a organização da sociedade civil Sou da Paz, conduzimos um experimento de campo direcionado a grupos ideologicamente distintos, incluindo Progressistas, Conservadores Cristãos, Empreendedores e pessoas de baixa renda. Cada grupo recebeu mensagens impulsionadas sobre uso de câmeras corporais por policiais. Nossos resultados indicam que a estruturação moral das mensagens aumentou significativamente o engajamento, com segmentos conservadores apresentando até 30% mais engajamento do que progressistas, que receberam mensagens neutras. Esses resultados contribuem para os campos da comunicação política, demonstrando como a segmentação ideológica e os apelos morais podem intensificar o engajamento online no Sul Global. O estudo oferece insights práticos para organizações da sociedade civil que buscam mobilizar apoiadores e influenciar o discurso público, ampliando a aplicação da MFT para novos contextos sociopolíticos e geográficos.

Esta nota de pesquisa argumenta que os avanços recentes na metodologia de inferência causal — notadamente a Revolução da Credibilidade nos métodos quantitativos e o desenvolvimento da inferência Bayesiana — oferecem um novo enquadramento para avaliar o rigor e a validade da pesquisa qualitativa em ciência política. Em vez de assumir que métodos qualitativos são inerentemente limitados em sua capacidade de generalização, propomos uma distinção mais precisa entre identificação causal e inferência estatística. Com base em dois desenvolvimentos recentes — o process tracing Bayesiano e o modelo de queries causais — mostramos que estudos de pequeno-n podem produzir inferências causais robustas, desde que as suposições de identificação sejam plausíveis, transparentes e avaliáveis. O objetivo da nota é sistematizar, de forma acessível e didática, os desafios e potenciais desses métodos, contribuindo para sua aplicação crítica no ensino e na pesquisa em ciência política.

This study analyzes the 2006 Bolivia-Brazil gas negotiations using supermodular game theory and leveraging unprecedented analysis of diplomatic cables. Our puzzle, from a rationalist perspective, is how a small power can prevail over a regional leader. While traditional theories suggest interdependence fosters cooperation, this case reveals it can harden positions and intensify bargaining crises when external actors intervene. By modeling strategic interactions between Bolivia, Brazil, and Venezuela, the study highlights how Venezuelan support strengthened Bolivia’s stance. The nationalization of hydrocarbons constrained Brazil’s response, as Lula faced domestic political limitations. The findings challenge conventional views by demonstrating that interdependence can trigger crises. The supermodular model contributes to the literature on international negotiations and energy diplomacy by providing a formalized approach to understanding bargaining rigidity in resource-driven disputes.

Does local oversight improve public service delivery? We study the effect of a mobile phone application that allows citizens to monitor school construction projects in Brazilian municipalities. The app prompts users to submit data about construction sites, sends such crowdsourced information to independent engineers, and contacts the mayors’ offices about project delays. Our results show that the app has a null impact on school construction indicators. Additionally, we find that politicians are unresponsive to individual requests. The results question the impact of bottom-up monitoring on public service performance and suggest that interventions targeted at other groups, or focused on different issues, may produce better policy outcomes.

This project investigates whether the International Liberal Order (ILO) is in substantive erosion or structural reorganization. The core argument is that this debate remains inconclusive because we still lack systematic empirical measures of the ILO over time. To address this gap, the project combines four strategies: (1) estimation of a latent measure of support for the ILO using dynamic IRT models applied to treaty ratification across multiple domains; (2) analysis of the substantive content of treaties with graded IRT and LLM-assisted coding; (3) measurement of trade openness through a two-dimensional typology based on tariff levels and universality (MFN vs. preferential); and (4) development of a formal game-theoretic model to microfound state decisions to join international commitments. The expected contribution is to provide a comparable quantitative measure of the ILO, distinguish formal from substantive adherence, and evaluate, with greater precision, systemic-level dynamics of transformation in the international order over time.