Abstract

Many researchers apply classical statistical decision theory to evaluate treatment choices and learn optimal policies. However, because this framework relies solely on realized outcomes under chosen actions and ignores counterfactuals, it cannot assess the quality of a decision relative to feasible alternatives at the unit level, which is an important requirement in some settings. For example, in pretrial bail decisions, a judge must balance crime prevention upon release against the risk of imposing unnecessary burdens on arrestees. A central challenge in this framework is identification: since only one potential outcome is observed per unit, counterfactual risk is typically not identifiable. We show that, under strong ignorability, counterfactual risk is identifiable if and only if the loss is additive in the potential outcomes. We further demonstrate that additive counterfactual losses can yield treatment recommendations that differ from those based on standard losses when more than two treatment options are available. We show that additive counterfactual losses capture not only decision accuracy but also decision difficulty, whereas standard losses reflect accuracy alone. Finally, we introduce a symbolic linear inverse program that determines whether a given counterfactual loss yields an identifiable risk, without requiring data.

Software

aihuman: Experimental Evaluation of Algorithm-Assisted Human Decision-Making — CRAN / GitHub

Related Papers

Zhou, Lingxiao, Kosuke Imai, Jason Lyall, and Georgia Papadogeorgou (2026). ``Dynamic Policy Evaluation and Learning with Spatio-temporal Data.''

Jia, Zeyang, Eli Ben-Michael, and Kosuke Imai (2026). ``Bayesian Safe Policy Learning with Chance Constrained Optimization: Application to Military Security Assessment during the Vietnam War.'' Journal of the Royal Statistical Society, Series A (Statistics in Society), Vol. 189, No. 3, pp. 1448-1472.

Imai, Kosuke, Sooahn Shin, D. James Greiner, and Ryan Halen (2026). ``Triage Score: A Counterfactual Risk Assessment Instrument.''

Koch, Benedikt, Kosuke Imai, and Tomasz Strzalecki (2026). ``An Axiomatic Foundation for Decisions with Counterfactual Utility.''

Li, Michael Lingzhi and Kosuke Imai (2025). ``Statistical Performance Guarantee for Subgroup Identification with Generic Machine Learning.''

Ben-Michael, Eli, D. James Greiner, Kosuke Imai, and Zhichao Jiang (2025). ``Safe Policy Learning through Extrapolation: Application to Pre-trial Risk Assessment.'' Journal of the American Statistical Association, Vol. 120, No. 551, pp. 1386-1399.

Ben-Michael, Eli, D. James Greiner, Melody Huang, Kosuke Imai, Zhichao Jiang, Sooahn Shin (2025). ``Does AI help humans make better decisions? A statistical evaluation framework for experimental and observational studies.'' Proceedings of the National Academy of Sciences, Vol. 122, No. 38, e2505106122.

Imai, Kosuke and Michael Lingzhi Li (2025). ``Statistical Inference for Heterogeneous Treatment Effects Discovered by Generic Machine Learning in Randomized Experiments.'' Journal of Business & Economic Statistics, Vol. 43, No. 1, pp. 256-268.

Jia, Zeyang, Kosuke Imai, and Michael Lingzhi Li (2025). ``Cramming Contextual Bandits for On-policy Statistical Evaluation.''

Zhang, Yi and Kosuke Imai (2025). ``Individualized Policy Evaluation and Learning under Clustered Network Interference.''

Zhang, Yi, Eli Ben-Michael, and Kosuke Imai (2024). ``Safe Policy Learning under Regression Discontinuity Designs.''

Ben-Michael, Eli, Kosuke Imai, and Zhichao Jiang (2024). ``Policy Learning with Asymmetric Counterfactual Utilities.'' Journal of the American Statistical Association, Vol. 119, No. 548, pp. 3045-3058.

Li, Michael Lingzhi and Kosuke Imai (2024). ``Neyman Meets Causal Machine Learning: Experimental Evaluation of Individualized Treatment Rules.'' Journal of Causal Inference, Vol 12, No. 1, pp. 1-20. Special Issue on Neyman (1923) and its influences on causal inference

Imai, Kosuke and Zhichao Jiang (2023). ``Principal Fairness for Human and Algorithmic Decision-Making.'' Statistical Science, Vol. 38, No. 2 (July), pp317-328.

Imai, Kosuke and Michael Lingzhi Li (2023). ``Experimental Evaluation of Individualized Treatment Rules.'' Journal of the American Statistical Association, Vol. 118, No. 541, pp. 242-256.

Imai, Kosuke, Zhichao Jiang, D. James Greiner, Ryan Halen, and Sooahn Shin (2023). ``Experimental Evaluation of Algorithm-Assisted Human Decision-Making: Application to Pretrial Public Safety Assessment.'' (with discussion) Journal of the Royal Statistical Society, Series A (Statistics in Society), Vol. 186, No. 2 (April), pp. 167-189. Read before the Royal Statistical Society.

Imai, Kosuke, and Aaron Strauss (2011). ``Estimation of Heterogeneous Treatment Effects from Randomized Experiments, with Application to the Optimal Planning of the Get-out-the-vote Campaign.'' Political Analysis, Vol. 19, No. 1 (Winter), pp. 1-19. Winner of Political Analysis Editors' Choice Award.