Orthogonal Policy Learning with Ordinal Outcomes
Abstract
Policy learning methods based on conditional average treatment effects can obscure subpopulation heterogeneity when applied to ordinal outcomes. We develop a policy learning framework for ordinal outcomes with heterogeneous utilities for individuals who strictly benefit from treatment and those who do not. Since the probability of strict benefit is only partially identified without further conditions, we adopt a minimax strategy which minimizes the worst-case regret over the identification region. To estimate the resulting nonsmooth objective, we combine smooth approximations with Neyman orthogonalization to remove first-order bias from nuisance estimation. We also derive an excess worst-case regret bound over a restricted policy class. The proposed method is validated through extensive simulations and an application to the 2022 Survey of Income and Program Participation (SIPP) data.
Related
- Is the treatment beneficial? Causal Inference with Ordinal Outcomes: A Bayesian Solution
- Doubly Robust Estimation and Inference for Longitudinal Win Probabilities with Ordinal Outcomes in Observational Studies
- Generalized Odds for Ordinal Outcomes: A Causal Inference Framework Adjusting for Covariates
- How Wrong Can a Rank-Based Sample-Size Calculation Be? A Sharp Bound of 16/9 for Ordinal Outcomes
- The risks of dichotomising ordinal outcomes: A spatial analysis of self-rated health in Western Europe
BibTeX
@misc{zhang2026orthogonal,
title = {{Orthogonal Policy Learning with Ordinal Outcomes}},
author = {Yue Zhang and Shanshan Luo and Yangbo He},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.22637},
eprint = {2609.22637},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.22637},
}
RIS
TY - UNPB
TI - Orthogonal Policy Learning with Ordinal Outcomes
AU - Yue Zhang
AU - Shanshan Luo
AU - Yangbo He
PY - 2026
DA - 2026/09/18
PB - arXiv
DO - 10.48550/arxiv.2609.22637
UR - https://arxiv.org/abs/2609.22637
AB - Policy learning methods based on conditional average treatment effects can obscure subpopulation heterogeneity when applied to ordinal outcomes. We develop a policy learning framework for ordinal outcomes with heterogeneous utilities for individuals who strictly benefit from treatment and those who do not. Since the probability of strict benefit is only partially identified without further conditions, we adopt a minimax strategy which minimizes the worst-case regret over the identification region. To estimate the resulting nonsmooth objective, we combine smooth approximations with Neyman orthogonalization to remove first-order bias from nuisance estimation. We also derive an excess worst-case regret bound over a restricted policy class. The proposed method is validated through extensive simulations and an application to the 2022 Survey of Income and Program Participation (SIPP) data.
ER -