Rerandomization to improve covariate balance in experiments

Kari Lock Morgan, Donald B. Rubin

Research output: Contribution to journalArticle

51 Citations (Scopus)

Abstract

Randomized experiments are the "gold standard" for estimating causal effects, yet often in practice, chance imbalances exist in covariate distributions between treatment groups. If covariate data are available before units are exposed to treatments, these chance imbalances can be mitigated by first checking covariate balance before the physical experiment takes place. Provided a precise definition of imbalance has been specified in advance, unbalanced randomizations can be discarded, followed by a rerandomization, and this process can continue until a randomization yielding balance according to the definition is achieved. By improving covariate balance, rerandomization provides more precise and trustworthy estimates of treatment effects.

Original languageEnglish (US)
Pages (from-to)1263-1282
Number of pages20
JournalAnnals of Statistics
Volume40
Issue number2
DOIs
StatePublished - Apr 1 2012

Fingerprint

Covariates
Randomisation
Experiment
Randomized Experiments
Causal Effect
Treatment Effects
Gold
Continue
Unit
Estimate
Imbalance
Randomization

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Morgan, Kari Lock ; Rubin, Donald B. / Rerandomization to improve covariate balance in experiments. In: Annals of Statistics. 2012 ; Vol. 40, No. 2. pp. 1263-1282.
@article{4fe5bd62de8445ca9fe26a263f24309a,
title = "Rerandomization to improve covariate balance in experiments",
abstract = "Randomized experiments are the {"}gold standard{"} for estimating causal effects, yet often in practice, chance imbalances exist in covariate distributions between treatment groups. If covariate data are available before units are exposed to treatments, these chance imbalances can be mitigated by first checking covariate balance before the physical experiment takes place. Provided a precise definition of imbalance has been specified in advance, unbalanced randomizations can be discarded, followed by a rerandomization, and this process can continue until a randomization yielding balance according to the definition is achieved. By improving covariate balance, rerandomization provides more precise and trustworthy estimates of treatment effects.",
author = "Morgan, {Kari Lock} and Rubin, {Donald B.}",
year = "2012",
month = "4",
day = "1",
doi = "10.1214/12-AOS1008",
language = "English (US)",
volume = "40",
pages = "1263--1282",
journal = "Annals of Statistics",
issn = "0090-5364",
publisher = "Institute of Mathematical Statistics",
number = "2",

}

Rerandomization to improve covariate balance in experiments. / Morgan, Kari Lock; Rubin, Donald B.

In: Annals of Statistics, Vol. 40, No. 2, 01.04.2012, p. 1263-1282.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Rerandomization to improve covariate balance in experiments

AU - Morgan, Kari Lock

AU - Rubin, Donald B.

PY - 2012/4/1

Y1 - 2012/4/1

N2 - Randomized experiments are the "gold standard" for estimating causal effects, yet often in practice, chance imbalances exist in covariate distributions between treatment groups. If covariate data are available before units are exposed to treatments, these chance imbalances can be mitigated by first checking covariate balance before the physical experiment takes place. Provided a precise definition of imbalance has been specified in advance, unbalanced randomizations can be discarded, followed by a rerandomization, and this process can continue until a randomization yielding balance according to the definition is achieved. By improving covariate balance, rerandomization provides more precise and trustworthy estimates of treatment effects.

AB - Randomized experiments are the "gold standard" for estimating causal effects, yet often in practice, chance imbalances exist in covariate distributions between treatment groups. If covariate data are available before units are exposed to treatments, these chance imbalances can be mitigated by first checking covariate balance before the physical experiment takes place. Provided a precise definition of imbalance has been specified in advance, unbalanced randomizations can be discarded, followed by a rerandomization, and this process can continue until a randomization yielding balance according to the definition is achieved. By improving covariate balance, rerandomization provides more precise and trustworthy estimates of treatment effects.

UR - http://www.scopus.com/inward/record.url?scp=84872006253&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84872006253&partnerID=8YFLogxK

U2 - 10.1214/12-AOS1008

DO - 10.1214/12-AOS1008

M3 - Article

AN - SCOPUS:84872006253

VL - 40

SP - 1263

EP - 1282

JO - Annals of Statistics

JF - Annals of Statistics

SN - 0090-5364

IS - 2

ER -