Rerandomization under Interference
Abstract
Covariates are widely used in randomized experiments to improve precision. However, in the presence of interference, where outcomes may depend on the treatment assignments of other units, standard covariate adjustment methods may fail to preserve desirable properties such as the no-harm property, meaning that incorporating covariates does not worsen estimator performance. Existing approaches that incorporate covariates under interference typically rely on specifying a particular interference structure, and their guarantees can be sensitive to misspecification. In this paper, we study how to incorporate covariate information in a way that preserves the no-harm property while remaining largely agnostic to the underlying interference structure. We focus on the estimation of the expected average treatment effect (EATE) using the Hájek estimator and study rerandomization under interference, a design-stage procedure that restricts the assignment space to allocations with sufficiently small covariate imbalance. Our framework also allows the covariates used for rerandomization to depend on the treatment assignment itself, such as the proportion of treated neighbors, which naturally arises in settings with interference. We show that, even under interference, rerandomization can improve estimation precision asymptotically relative to unrestricted Bernoulli randomization, relying only on mild conditions on the dependence structure across units. When a conservative dependence graph is available, we further develop an optimization-based conservative variance estimator for inference under rerandomization.
Related
- Rerandomization and high-dimensional adjustment with assignment-dependent auxiliary variables under network interference
- Data-Adaptive Rerandomization for 2K Factorial Designs
- Spatial causal inference in the presence of unmeasured confounding and interference
- Selective Interference Suppression of Siamese-Net in Heterogeneous Interference Channels
- Interference-Free Capacity of a Binary Interference Channel with Causal Observation
BibTeX
@misc{yuan2026rerandomization,
title = {{Rerandomization under Interference}},
author = {Ziang Yuan and Xinran Li and Shuangning Li},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.33730},
eprint = {2609.33730},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.33730},
}
RIS
TY - UNPB
TI - Rerandomization under Interference
AU - Ziang Yuan
AU - Xinran Li
AU - Shuangning Li
PY - 2026
DA - 2026/09/27
PB - arXiv
DO - 10.48550/arxiv.2609.33730
UR - https://arxiv.org/abs/2609.33730
AB - Covariates are widely used in randomized experiments to improve precision. However, in the presence of interference, where outcomes may depend on the treatment assignments of other units, standard covariate adjustment methods may fail to preserve desirable properties such as the no-harm property, meaning that incorporating covariates does not worsen estimator performance. Existing approaches that incorporate covariates under interference typically rely on specifying a particular interference structure, and their guarantees can be sensitive to misspecification. In this paper, we study how to incorporate covariate information in a way that preserves the no-harm property while remaining largely agnostic to the underlying interference structure. We focus on the estimation of the expected average treatment effect (EATE) using the Hájek estimator and study rerandomization under interference, a design-stage procedure that restricts the assignment space to allocations with sufficiently small covariate imbalance. Our framework also allows the covariates used for rerandomization to depend on the treatment assignment itself, such as the proportion of treated neighbors, which naturally arises in settings with interference. We show that, even under interference, rerandomization can improve estimation precision asymptotically relative to unrestricted Bernoulli randomization, relying only on mild conditions on the dependence structure across units. When a conservative dependence graph is available, we further develop an optimization-based conservative variance estimator for inference under rerandomization.
ER -