Peer-Reviewed Publications

Andrew S. Rosenberg, "Measuring Racial Bias in International Migration Flows." International Studies Quarterly. Forthcoming.

Abstract: Are international migration flows racially biased? Despite widespread consensus that racism and xenophobia affect migration processes, no measure exists to provide systematic evidence on this score. In this short paper, I construct such a measure—the migration deviation. Migration deviations are the difference between the observed migration between states, and the flow that we would predict based on a racially blind model that includes a wide variety of political and economic factors. Using this measure, I conduct a descriptive analysis and provide evidence that migrants from majority black states migrate far less than we would expect under a racially blind model. These results pave a new way for scholars to study international racial inequality.

PDF.     Replication Materials.

Andrew S. Rosenberg, Austin J. Knuppe, and Bear F. Braumoeller, "Unifying the Study of Asymmetric Hypotheses" Political Analysis 25(3): 381--401. DOI:

Abstract: This article presents a conceptual clarification of asymmetric hypotheses and a discussion of methodologies available to test them. Despite the existence of a litany of theories that posit asymmetric causation, most empirical studies fail to capture their core insight: boundaries separating zones of data from areas that lack data are substantively interesting. We discuss existing set-theoretic and large-N approaches to the study of asymmetric hypotheses, evaluate their relative merits, and give three examples of how asymmetric hypotheses can be studied with this suite of tools. 

PDF.     Publisher Version.     Replication Materials.     BibTeX.     Software.


Working Papers (Drafts available upon request)

William Minozzi, Andrew S. Rosenberg, Eun Bin Chung, and Matthew P. Hitt, "It Takes Two: Heterogeneous Motivations for Reputations in International Relations." Under Review

Abstract: Most scholarship on reputations in IR argues either that states have instrumental motives, or that they cultivate reputations because of intrinsic reasons—or that the two accounts are observationally equivalent. Such equivalence suggests that motivational heterogeneity does not matter. In contrast, we claim that this equivalence results from a failure to identify a missing counterfactual: behavior in a system with limited capacity for reputations. We develop a theoretical minimal reputation environment to explain how actors with different motivations behave within various institutions, and test our predictions using an incentivized experiment based on trust games. Categorizing participants based on “folk realism,” we find that consequence-oriented realists respond dramatically to reputation enhancing institutions, leapfrogging appropriateness-oriented idealists in trust and trustworthiness. Moreover, a leader-follower dynamic emerges, as realists respond quickly to novel institutions and idealists adapt more slowly. Ultimately, we argue that it takes two kinds of motivations to explain reputations in IR.

Jan H. Pierskalla, Adam Lauretig, Andrew S. Rosenberg, and Audrey Sacks, "Democratization and the Civil Service---An Analysis of Promotion Patterns in Indonesia.'' Under Review

Abstract: What is the effect of democratization on meritocratic practices in the civil service? Democratization increases performance incentives within the bureaucracy. This leads to more meritocracy for individuals with performance-enhancing characteristics, e.g., educational attainment, that cross-cut political cleavages. When politicized cleavages align with performance-enhancing characteristics of civil servants, democratization increases discrimination. We test this argument using administrative data from Indonesia. Our data covers the full universe of career histories of all 4+ million currently active civil servants. We exploit the exogenous timing of Indonesia's democratization in 1999, paired with an individual-level panel data design, for identification purposes. We find strong evidence that democratization amplified the positive effects of educational attainment on career advancement but simultaneously worsened the career prospects of female and religious minority civil servants. We replicate these patterns for the staggered introduction of direct elections at the district government level. The gender and religious minority penalties are strongest for promotions at the lowest rungs of the administrative hierarchy and for employees of departments under the leadership of conservative Muslim parties. These penalties are not alleviated by increased female leadership in the bureaucracy. 

Andrew S. Rosenberg, "Emigration Underflows and Economic Development." Under Review

Abstract: The conventional wisdom among scholars of international migration is that emigration either harms a sending state's economic growth via a ``brain drain,'' or that it produces incentives for citizens to acquire education and generates a ``brain gain.'' I argue that this debate results from a failure to identify a missing counterfactual: the world in which emigration rates match neoclassical economic expectations. Comparing observed emigration to a counterfactual prediction allows one to infer if a state has more or less emigration than expected. I create a model of this counterfactual and develop a theory of a negative incentive effect to explain why some states suffer a ``drain'' while others experience a ``gain.'' Using a novel dataset, I find that underflows in emigration are associated with less human capital accumulation and decreased economic growth. This result holds strongest for states in the developing world, suggesting an unexplored dimension of international inequality. 

Andrew S. Rosenberg, "Causal Probabilities and Forensic Social Science." 

Abstract: Social science is a forensic discipline because most phenomena of interest have occurred and the researcher's task is to adjudicate among many causes of these events. However, the debate about the best way to test complex causal relationships is not discerning and falls along familiar quantitative/qualitative lines. To remedy this situation I present two probabilistic causal inference estimators, the probability of necessary and sufficient causation, that provide a unified framework for testing these relationships. I use a novel simulation study to show that these estimators strictly dominate the inferences made with existing large-N and set-theoretic techniques.

Andrew S. Rosenberg, William Minozzi, Elias Assaf, and Christopher Gelpi, "Man, the State, and War: An Experimental Approach."

Abstract: Both scholars and interested observers of international politics share the intuitive sense that “leaders matter” in determining the onset and prevention of war. Much of the international relations literature on this question, however, has struggled to identify systematic causal effects. The difficulty in identifying the causal impact of leaders on the outbreak of war is primarily rooted in two issues: 1) the measurement of individual characteristics at a distance, and 2) the confounding of leadership characteristics with other national and international covariates in observational data. We propose a method for overcoming these problems through the online experimental analysis of crisis bargaining. We construct an online simulacrum for crisis bargaining that captures many of the key features of this literature while avoiding some of the important shortcomings of typical behavioral economics games for the study of war. Game instructions and screen shots of gameplay are included in our appendices. Next, we identify a key individual leadership characteristic - narcissism - and theorize its expected impact on play in our game. We then conduct computer simulations of game play based upon randomized assignment of narcissistic and non-narcissistic leaders to different simulated international systems. Our analysis of the simulated data illustrates the viability of our approach and provides hypotheses for the impact of leadership narcissism at the system level. At the same time, our results also highlight the importance of careful theorizing about causal process when examining treatment effects in a complex and strategic context.