Sebastian Galiani | Searcharxiv

arXiv2026

Does generative AI narrow education-based productivity gaps? Evidence from a randomized experiment

Does generative artificial intelligence (AI) widen or narrow productivity gaps across workers? We study this in a randomized online experiment with 1,174 adults aged 25-45 who completed a workplace-style problem-solving task with or without a generative AI assistant, followed by an unassisted module. AI improves performance for all participants, but gains are larger among those with less education. Without AI, higher-education participants outperform lower-education participants by 0.548 standard deviations; with AI, the gap falls to 0.139, closing about three-quarters of the initial difference. Chat logs show that lower-education participants obtain substantial assistance, while higher-education participants use AI more effectively. Gains are not purely due to delegation: treated participants do not perform worse once AI is removed, and lower-education participants retain part of their improvement, although a sizable gap re-emerges. Intensive AI use raises assisted performance regardless of participants' own effort, but follow-up performance improves only when intensive use is combined with sustained effort. Generative AI narrows effective productivity differences in task execution, while human-capital differences continue to shape unassisted performance and tool use.

econ.GN↗

arXiv2025

Beyond Bonferroni: Hierarchical Multiple Testing in Empirical Research

Empirical research in the social and medical sciences frequently involves testing multiple hypotheses simultaneously, increasing the risk of false positives due to chance. Classical multiple testing procedures, such as the Bonferroni correction, control the family-wise error rate (FWER) but tend to be overly conservative, reducing statistical power. Stepwise alternatives like the Holm and Hochberg procedures offer improved power while maintaining error control under certain dependence structures. However, these standard approaches typically ignore hierarchical relationships among hypotheses -- structures that are common in settings such as clinical trials and program evaluations, where outcomes are often logically or causally linked. Hierarchical multiple testing procedures -- including fixed sequence, fallback, and gatekeeping methods -- explicitly incorporate these relationships, providing more powerful and interpretable frameworks for inference. This paper reviews key hierarchical methods, compares their statistical properties and practical trade-offs, and discusses implications for applied empirical research.

econ.EM↗