About

Adaner Usmani is an Assistant Professor of Sociology and Social Studies at Harvard University. He obtained his Ph.D. in Sociology from New York University in 2017 and was a Postdoctoral Fellow at the Watson Institute at Brown University from 2017 to 2019. His research is driven by questions about the distribution of flourishing and suffering in modern societies, with a particular interest in the relationship between normative commitments and empirical work. Most of his ongoing research focuses on American mass incarceration, including projects with John Clegg and Chris Lewis that aim to explain American punishment through comparative and historical analysis. His work involves collecting data on prisons, policing, and penal spending, and he employs computational and quantitative methods, which he has also taught at the graduate level. Despite his focus on quantitative approaches, he maintains a strong interest in the comparative historical method and qualitative social science more generally.

Research topics

• Sociology
• Mathematics
• Machine Learning
• Computer Science
• Artificial Intelligence
• Statistics
• Engineering
• Econometrics
• Mathematical economics
• Geography
• Mathematical analysis
• Medicine
• Positive economics
• Neoclassical economics
• Epistemology
• Philosophy
• Economics

Selected publications

• How Segregation Ruins Inference: A Sociological Simulation of the Inequality Equilibrium

  Social Forces · 2024 · 12 citations

  Senior authorCorresponding
  • Sociology
  • Sociology
  • Mathematical economics

  Abstract Why do many people underestimate economic and racial inequality and maintain that theirs is a meritocratic society? Existing work suggests that people are rationalizing, misinformed, or misled. This article proposes an additional explanation: Inequality itself makes economic and racial disparities difficult to understand. In unequal societies, individuals establish their networks at formative institutions patterned by class and race. As a result, they unwittingly condition on key causal pathways when making descriptive and causal inferences about inequality. We use a simple agent-based model to show that, under circumstances typical to highly stratified societies, individuals will underestimate the extent of economic and racial inequality, downplay the importance of inherited advantages, and overestimate the relative importance of individual ability. Moreover, we show that they will both underestimate the extent of racial discrimination and overestimate its relative importance. Because segregated social worlds bias inference in these ways, all individuals (rich and poor) have principled reasons to favor less redistribution than they would if their social worlds were more integrated.

  DOI

• Labor markets and incarceration: The China shock to American punishment

  Criminology · 2023 · 3 citations

  • Political Science
  • Economics
  • Demographic economics

  Abstract Studies have failed to show a positive effect of unemployment on incarceration despite reasons to expect such a relationship. We note that prior estimates have been muddied by the absence of substate data, a focus on prisons rather than on jails, limited measures of unemployment, and the fact that the health of the labor market is endogenous to incarceration. We instrument for local exposure to the rise of Chinese exports (“the China Shock”) to estimate the effect of job loss on American incarceration. Marshaling a new data set of prisoners and jail inmates by race at the commuting zone level, we show that negative shocks to local labor markets led to significant increases in total incarceration rates for both Blacks and Whites. The effect seems to be driven by increased prison rather than jail populations. This estimate is invisible to ordinary least squares, which may help explain null results reported by past work. Counterfactual exercises suggest that the effect of job loss was punitively consequential. Had employment gains from the 1990s been preserved into the 2000s, the U.S. incarceration rate would have grown significantly less than it did.

  PublisherDOI

• Measuring the predictability of life outcomes with a scientific mass collaboration

  Proceedings of the National Academy of Sciences · 2020 · 319 citations

  • Computer Science
  • Machine Learning
  • Artificial Intelligence

  How predictable are life trajectories? We investigated this question with a scientific mass collaboration using the common task method; 160 teams built predictive models for six life outcomes using data from the Fragile Families and Child Wellbeing Study, a high-quality birth cohort study. Despite using a rich dataset and applying machine-learning methods optimized for prediction, the best predictions were not very accurate and were only slightly better than those from a simple benchmark model. Within each outcome, prediction error was strongly associated with the family being predicted and weakly associated with the technique used to generate the prediction. Overall, these results suggest practical limits to the predictability of life outcomes in some settings and illustrate the value of mass collaborations in the social sciences.

  PublisherOA PDFDOI

Frequent coauthors

• Antje Kirchner

  University of Nebraska–Lincoln

  5 shared

• Kivan Polimis

  University of Washington

  3 shared

• Ridhi Kashyap

  University of Oxford

  3 shared

• Connor Gilroy

  University of Washington

  3 shared

• Kirstie Whitaker

  Turing Institute

  3 shared

• Anna Filippova

  Lomonosov Moscow State University

  3 shared

• Allison C. Morgan

  University of Colorado Boulder

  3 shared

• Onur Varol

  Sabancı Üniversitesi

  2 shared

Similar researchers at Harvard University

• Walter C Willetth-429
• JoAnn E Mansonh-359
• Randy L. Bucknerh-268
• Edward Giovannuccih-267
• Kristen Andersonh-263