The Illusion of Learning from Observational Data: An Empirical Bayes Perspective

Bohan Wu, Sebastian Salazar, Donald P. Green, David M. Blei · 2026

Read The Illusion of Learning from Observational Data: An Empirical Bayes Perspective by Bohan Wu, Sebastian Salazar, Donald P. Green, David M. Blei (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-08853.txt.

Description

Wu and three colleagues study a trap in observational research, where studies without randomization can look increasingly convincing even as hidden bias persists. Building on work from 2004 on this illusion of learning, they frame the problem with empirical Bayes shrinkage. That method pulls noisy estimates toward a shared prior. They prove that with enough calibration and observation studies, both the bias pattern and the true effect can be recovered consistently. They test the method in simulations and a semi-synthetic water-use experiment.

Catalog record

CREATORBohan Wu, Sebastian Salazar, Donald P. Green, David M. Blei
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE106331
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.08853

Browse the full library · About the Tower of Babel