Inverse Probability Weighting in a Post-Bayesian World

Owen Thomas, William Denault, Valeria Vitelli · 2026

Read Inverse Probability Weighting in a Post-Bayesian World by Owen Thomas, William Denault, Valeria Vitelli (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-2606-28685.txt.

Description

Owen Thomas, William Denault, and Valeria Vitelli rescue a workhorse correction from a framework clash. Inverse probability weighting fixes selection bias in frequentist statistics by upweighting groups that show up too rarely. Bayesian analysis with possibly misspecified models has been moving toward a post-Bayesian stance that replaces the likelihood with a divergence, and weighting had no proper home there.

Thomas and colleagues recast the correction as a reweighting of the Kullback-Leibler divergence between the model and the true value behind the data, and prove convergence and traits of the resulting posteriors. Two simulated examples under selection bias show the method working. A large registry study, using prostate-specific antigen readings to predict prostate cancer deaths amid registry biases, shows it at real scale. Problems once out of reach inside a Bayesian approach become routine.

Catalog record

CREATOROwen Thomas, William Denault, Valeria Vitelli
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE66 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.28685

Browse the full library · About the Tower of Babel