Uncovering College Effect Heterogeneity

The Institute for Research on Poverty hosts seminars on the UW-Madison campus most Thursdays during the academic year. These presentations are free and open to the public. Room 8417.

media release: Speaker: Jennie Brand, professor, Department of Sociology, UCLA

12:15–1:30 pm, 8417 Sewell Social Science Building

Co-sponsored with the Department of Sociology

This talk explores recent advances in causal inference and machine learning with an emphasis on identifying and interpreting treatment effect heterogeneity. Using an empirical application on the heterogeneous effects of completing a 4-year college degree in the United States on low-wage work and poverty across the career, I illustrate how causal machine-learning methods can be applied to substantively important questions. Understanding effect heterogeneity is central to evaluating how scarce resources are distributed, improving treatment assignment to enhance average outcomes, and assessing the external validity of findings beyond the study population. The talk highlights propensity- and covariate-based partitioning strategies, as well as tree-based methods, for detecting and characterizing heterogeneous treatment effects. By integrating these approaches, the analysis reexamines and extends established findings on educational inequality in reducing socioeconomic disadvantage.