Bias in Balance Optimization Subset Selection

Exploration through examples

Hee Youn Kwon, Jason J. Sauppe, Sheldon Howard Jacobson

Research output: Contribution to journalArticle

Abstract

When estimating a treatment effect from observational data, researchers encounter bias regardless of estimation methods. In this paper, we focus on a particular method of estimation called Balance Optimization Subset Selection (BOSS). This paper investigates all the possible cases that may lead to bias in the context of BOSS, provides examples for those cases and tries to mitigate the bias. While doing so, we define a balance hierarchy and a correct imbalance measure which corresponds to the form of the response functions. In addition, new imbalance measures drawn from the Cramer-von Mises test statistic are introduced. The cases of insufficient data and suboptimality that can arise in causal analysis with BOSS are also presented.

Original languageEnglish (US)
Pages (from-to)67-80
Number of pages14
JournalJournal of the Operational Research Society
Volume70
Issue number1
DOIs
StatePublished - Jan 2 2019

Fingerprint

Set theory
Statistics
Imbalance

Keywords

  • Causal analysis
  • optimisation
  • subset selection

ASJC Scopus subject areas

  • Management Information Systems
  • Strategy and Management
  • Management Science and Operations Research
  • Marketing

Cite this

Bias in Balance Optimization Subset Selection : Exploration through examples. / Kwon, Hee Youn; Sauppe, Jason J.; Jacobson, Sheldon Howard.

In: Journal of the Operational Research Society, Vol. 70, No. 1, 02.01.2019, p. 67-80.

Research output: Contribution to journalArticle

@article{d66cc7bc3b1543d68a00d5e9b77bf3a8,
title = "Bias in Balance Optimization Subset Selection: Exploration through examples",
abstract = "When estimating a treatment effect from observational data, researchers encounter bias regardless of estimation methods. In this paper, we focus on a particular method of estimation called Balance Optimization Subset Selection (BOSS). This paper investigates all the possible cases that may lead to bias in the context of BOSS, provides examples for those cases and tries to mitigate the bias. While doing so, we define a balance hierarchy and a correct imbalance measure which corresponds to the form of the response functions. In addition, new imbalance measures drawn from the Cramer-von Mises test statistic are introduced. The cases of insufficient data and suboptimality that can arise in causal analysis with BOSS are also presented.",
keywords = "Causal analysis, optimisation, subset selection",
author = "Kwon, {Hee Youn} and Sauppe, {Jason J.} and Jacobson, {Sheldon Howard}",
year = "2019",
month = "1",
day = "2",
doi = "10.1080/01605682.2017.1421848",
language = "English (US)",
volume = "70",
pages = "67--80",
journal = "Journal of the Operational Research Society",
issn = "0160-5682",
publisher = "Palgrave Macmillan Ltd.",
number = "1",

}

TY - JOUR

T1 - Bias in Balance Optimization Subset Selection

T2 - Exploration through examples

AU - Kwon, Hee Youn

AU - Sauppe, Jason J.

AU - Jacobson, Sheldon Howard

PY - 2019/1/2

Y1 - 2019/1/2

N2 - When estimating a treatment effect from observational data, researchers encounter bias regardless of estimation methods. In this paper, we focus on a particular method of estimation called Balance Optimization Subset Selection (BOSS). This paper investigates all the possible cases that may lead to bias in the context of BOSS, provides examples for those cases and tries to mitigate the bias. While doing so, we define a balance hierarchy and a correct imbalance measure which corresponds to the form of the response functions. In addition, new imbalance measures drawn from the Cramer-von Mises test statistic are introduced. The cases of insufficient data and suboptimality that can arise in causal analysis with BOSS are also presented.

AB - When estimating a treatment effect from observational data, researchers encounter bias regardless of estimation methods. In this paper, we focus on a particular method of estimation called Balance Optimization Subset Selection (BOSS). This paper investigates all the possible cases that may lead to bias in the context of BOSS, provides examples for those cases and tries to mitigate the bias. While doing so, we define a balance hierarchy and a correct imbalance measure which corresponds to the form of the response functions. In addition, new imbalance measures drawn from the Cramer-von Mises test statistic are introduced. The cases of insufficient data and suboptimality that can arise in causal analysis with BOSS are also presented.

KW - Causal analysis

KW - optimisation

KW - subset selection

UR - http://www.scopus.com/inward/record.url?scp=85049018991&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85049018991&partnerID=8YFLogxK

U2 - 10.1080/01605682.2017.1421848

DO - 10.1080/01605682.2017.1421848

M3 - Article

VL - 70

SP - 67

EP - 80

JO - Journal of the Operational Research Society

JF - Journal of the Operational Research Society

SN - 0160-5682

IS - 1

ER -