Unbiased Regression-Adjusted Estimation of Average Treatment Effects in Randomized Controlled Trials

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Abadie, Alberto, Ghadiri, Mehrdad, Jadbabaie, Ali, JafariNodeh, Mahyar
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915985170104320
author Abadie, Alberto
Ghadiri, Mehrdad
Jadbabaie, Ali
JafariNodeh, Mahyar
author_facet Abadie, Alberto
Ghadiri, Mehrdad
Jadbabaie, Ali
JafariNodeh, Mahyar
contents This article introduces a leave-one-out regression adjustment (LOORA) for estimating average treatment effects in randomized controlled trials. In finite samples, LOORA removes the bias of conventional regression adjustment and yields exact variance formulas for regression-adjusted Horvitz-Thompson and difference-in-means estimators. Ridge regularization curbs the influence of high-leverage observations, improving stability and precision in small samples. In large samples, LOORA matches the variance of the regression-adjusted estimator in Lin (2013) while remaining exactly unbiased. Two within-subject experimental applications, each providing a realistic joint distribution of potential outcomes as ground truth, show that LOORA removes substantial bias and achieves confidence interval coverage close to the nominal level.
format Preprint
id arxiv_https___arxiv_org_abs_2511_03236
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Unbiased Regression-Adjusted Estimation of Average Treatment Effects in Randomized Controlled Trials
Abadie, Alberto
Ghadiri, Mehrdad
Jadbabaie, Ali
JafariNodeh, Mahyar
Econometrics
Methodology
62J07, 62P20
G.3
This article introduces a leave-one-out regression adjustment (LOORA) for estimating average treatment effects in randomized controlled trials. In finite samples, LOORA removes the bias of conventional regression adjustment and yields exact variance formulas for regression-adjusted Horvitz-Thompson and difference-in-means estimators. Ridge regularization curbs the influence of high-leverage observations, improving stability and precision in small samples. In large samples, LOORA matches the variance of the regression-adjusted estimator in Lin (2013) while remaining exactly unbiased. Two within-subject experimental applications, each providing a realistic joint distribution of potential outcomes as ground truth, show that LOORA removes substantial bias and achieves confidence interval coverage close to the nominal level.
title Unbiased Regression-Adjusted Estimation of Average Treatment Effects in Randomized Controlled Trials
topic Econometrics
Methodology
62J07, 62P20
G.3
url https://arxiv.org/abs/2511.03236