Breaking the Winner's Curse with Bayesian Hybrid Shrinkage

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Mudd, Richard, Friedberg, Rina, Gorbachev, Ilya, Nassif, Houssam, Zaidi, Abbas
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915607065133056
author Mudd, Richard
Friedberg, Rina
Gorbachev, Ilya
Nassif, Houssam
Zaidi, Abbas
author_facet Mudd, Richard
Friedberg, Rina
Gorbachev, Ilya
Nassif, Houssam
Zaidi, Abbas
contents A 'Winner's Curse' arises in large-scale online experimentation platforms when the same experiments are used to both select treatments and evaluate their effects. In these settings, classical difference-in-means estimators of treatment effects are upwardly biased and conventional confidence intervals are rendered invalid. The bias scales with the magnitude of sampling variability and the selection threshold, and inversely with the treatment's true effect size. We propose a new Bayesian approach that incorporates experiment-specific 'local shrinkage' factors that mitigate sensitivity to the choice of prior and improve robustness to assumption violations. We demonstrate how the associated posterior distribution can be estimated without numerical integration techniques, making it a practical choice for at-scale deployment. Through simulation, we evaluate the performance of our approach under various scenarios and find that it performs well even when assumptions about the sampling and selection processes are violated. In an empirical evaluation, our approach demonstrated superior performance over alternative methods, providing more accurate estimates with well-calibrated uncertainty quantification.
format Preprint
id arxiv_https___arxiv_org_abs_2511_06318
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Breaking the Winner's Curse with Bayesian Hybrid Shrinkage
Mudd, Richard
Friedberg, Rina
Gorbachev, Ilya
Nassif, Houssam
Zaidi, Abbas
Methodology
Applications
A 'Winner's Curse' arises in large-scale online experimentation platforms when the same experiments are used to both select treatments and evaluate their effects. In these settings, classical difference-in-means estimators of treatment effects are upwardly biased and conventional confidence intervals are rendered invalid. The bias scales with the magnitude of sampling variability and the selection threshold, and inversely with the treatment's true effect size. We propose a new Bayesian approach that incorporates experiment-specific 'local shrinkage' factors that mitigate sensitivity to the choice of prior and improve robustness to assumption violations. We demonstrate how the associated posterior distribution can be estimated without numerical integration techniques, making it a practical choice for at-scale deployment. Through simulation, we evaluate the performance of our approach under various scenarios and find that it performs well even when assumptions about the sampling and selection processes are violated. In an empirical evaluation, our approach demonstrated superior performance over alternative methods, providing more accurate estimates with well-calibrated uncertainty quantification.
title Breaking the Winner's Curse with Bayesian Hybrid Shrinkage
topic Methodology
Applications
url https://arxiv.org/abs/2511.06318