Ying Jin, Naoki Egami
arXiv 15 Sep 2026 · Statistics — Methodology
arXiv:2609.17296 · PDF · Extracted main text
Policy learning aims to determine who should be treated based on individual characteristics. In high-stakes settings such as medicine and public policy where safety is a central concern, improving the average outcomes alone may not be sufficient: decision makers may also seek to protect individuals from harm, in line with the Hippocratic principle of “do no harm.” In this paper, we propose conformal policy learning (CPL), a policy learning procedure with a new distribution-free safety guarantee that controls the probability of assigning treatment to an individual who would be harmed relative to control. CPL views each treatment decision as testing a hypothesis of counterfactual harm and assigns treatment by thresholding conformal p-values. These p-values use observable proxies and selective calibration to address the challenge that the potential outcomes under comparison are never simultaneously observed. For randomized experiments, under standard exchangeability conditions, CPL provides finite-sample safety guarantee at a user-specified level, without imposing any outcome modeling assumptions. Moreover, when the outcome model is consistently estimated, CPL achieves asymptotically optimal welfare subject to the safety constraint. In observational studies, CPL with learn-then-balance weights achieves doubly robust safety guarantees. We evaluate CPL through extensive simulations and apply it to an empirical study of AI-powered interventions designed to reduce conspiracy beliefs.
appendix boundary found by appendix_command · 38% of the source is main text. Read the extracted text to check this.
The works this paper leans on most, across its whole bibliography — not restricted to papers in our corpus. Ranked by composite intensity, which combines how often a work is mentioned, how many sections mention it, and how much of that falls in the main text rather than the appendix.
| Reference | Intensity | Mentions | Sections | Main text | |
|---|---|---|---|---|---|
| 1 | Athey, Susan and Wager, Stefan (2021) Policy learning with observational data | 1.000 | 5 | 3 | 100% |
| 2 | Kitagawa, Toru and Tetenov, Aleksey (2018) Who should be treated? empirical welfare maximization methods for treatment choice | 1.000 | 5 | 3 | 100% |
| 3 | Manski, Charles F (2004) Statistical Treatment Rules for Heterogeneous Populations | 1.000 | 5 | 3 | 100% |
| 4 | Jin, Ying and Candès, Emmanuel J (2023) Selection by prediction with conformal p-values self | 0.956 | 8 | 4 | 88% |
| 5 | Hirano, Keisuke and Porter, Jack R (2009) Asymptotics for Statistical Treatment Rules | 0.928 | 4 | 3 | 100% |
| 6 | Lei, Lihua and Candès, Emmanuel J (2021) Conformal inference of counterfactuals and individual treatment effects | 0.928 | 4 | 3 | 100% |
| 7 | Zhao, Yingqi and Zeng, Donglin and Rush, A John and Kosorok, Michael R (2012) Estimating Individualized Treatment Rules using Outcome Weighted Learning | 0.928 | 4 | 3 | 100% |
| 8 | Kallus, Nathan (2022) What's the harm? sharp bounds on the fraction negatively affected by treatment | 0.920 | 9 | 5 | 78% |
| 9 | Li, Haoxuan and Zheng, Chunyuan and Cao, Yixiao and Geng, Zhi and Li… (2023) Trustworthy policy learning under the counterfactual no-harm criterion | 0.855 | 8 | 5 | 62% |
| 10 | Vovk, Vladimir and Gammerman, Alexander and Shafer, Glenn (2005) Algorithmic learning in a random world | 0.843 | 4 | 4 | 75% |
Showing the top 10 of 57 scored citations.