When Does Inexact Matching Ensure Balance and Inference without Adjustment?
Abstract
One-to-one matching without replacement is a classical approach to constructing comparable treated and control samples in the design of observational studies. It pairs each treated unit with a distinct control while minimizing a covariate distance objective. With continuous covariates, the matched pairs generally remain inexact, which contributes to bias in downstream analysis. Its key theoretical properties, such as the resulting imbalance between matched pairs and when it is negligible to support valid inference, remain unclear. In this paper, we analyze one-to-one matching based on d-dimensional, continuous covariates with a quadratic covariate-distance objective. First, we find that when d≤ 3, under standard conditions on the propensity score ensuring abundant control samples near each treated sample, the imbalance (difference between within-group averages) is root-n negligible uniformly over the family of smooth functions with a common first- and second-order derivative bound. However, such balance is subject to a dimension restriction, as we construct examples in which the imbalance is root-n non-negligible when d=4 and dominates root-n rate when d>4. Second, we show that when d≤ 3, the matched design allows valid Wald-type and bootstrap inference for the average treatment effect on the treated, distributional treatment effects, and quantile treatment effects. Thus, the same outcome-blind matched design supports various downstream inferences without having to tailor the design to the targets. Finally, paired randomization inference based on the matched design is asymptotically valid in the super-population sense for d≤ 3 but can fail when d=4. We corroborate the theoretical results with numerical experiments.
Related
- Automated, Efficient, and Model-Free Inference for Randomized Clinical Trials via Data-Driven Covariate Adjustment
- Higher Order Asymptotics of Regression Adjustment Estimators for Causal Inference
- Covariate Adjustment for Time-Restricted Marginal Win Statistics in Randomized Trials: Assumption-Lean Inference and Guaranteed Efficiency Gain
- Exhaustive adjustment-set selection and causal effects for distribution-valued outcomes, with inference under latent Gaussian spatio-temporal dependence
- Nesting or resolution? Facility proxies for variance estimation versus confounding adjustment in claims data without facility identifiers
BibTeX
@misc{jin2026when,
title = {{When Does Inexact Matching Ensure Balance and Inference without Adjustment?}},
author = {Ying Jin},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.10873},
eprint = {2610.10873},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.10873},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - When Does Inexact Matching Ensure Balance and Inference without Adjustment?
AU - Ying Jin
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2610.10873
UR - https://arxiv.org/abs/2610.10873
AB - One-to-one matching without replacement is a classical approach to constructing comparable treated and control samples in the design of observational studies. It pairs each treated unit with a distinct control while minimizing a covariate distance objective. With continuous covariates, the matched pairs generally remain inexact, which contributes to bias in downstream analysis. Its key theoretical properties, such as the resulting imbalance between matched pairs and when it is negligible to support valid inference, remain unclear. In this paper, we analyze one-to-one matching based on d-dimensional, continuous covariates with a quadratic covariate-distance objective. First, we find that when d≤ 3, under standard conditions on the propensity score ensuring abundant control samples near each treated sample, the imbalance (difference between within-group averages) is root-n negligible uniformly over the family of smooth functions with a common first- and second-order derivative bound. However, such balance is subject to a dimension restriction, as we construct examples in which the imbalance is root-n non-negligible when d=4 and dominates root-n rate when d>4. Second, we show that when d≤ 3, the matched design allows valid Wald-type and bootstrap inference for the average treatment effect on the treated, distributional treatment effects, and quantile treatment effects. Thus, the same outcome-blind matched design supports various downstream inferences without having to tailor the design to the targets. Finally, paired randomization inference based on the matched design is asymptotically valid in the super-population sense for d≤ 3 but can fail when d=4. We corroborate the theoretical results with numerical experiments.
ER -