Assessing lack of common support in causal inference using bayesian nonparametrics: Implications for evaluating the effect of breastfeeding on children's cognitive outcomes

Jennifer Hill, Yu Sung Su

Research output: Contribution to journalArticle

Abstract

Causal inference in observational studies typically requires making comparisons between groups that are dissimilar. For instance, researchers investigating the role of a prolonged duration of breastfeeding on child outcomes may be forced to make comparisons between women with substantially different characteristics on average. In the extreme there may exist neighborhoods of the covariate space where there are not sufficient numbers of both groups of women (those who breastfed for prolonged periods and those who did not) to make inferences about those women. This is referred to as lack of common support. Problems can arise when we try to estimate causal effects for units that lack common support, thus we may want to avoid inference for such units. If ignorability is satisfied with respect to a set of potential confounders, then identifying whether, or for which units, the common support assumption holds is an empirical question. However, in the high-dimensional covariate space often required to satisfy ignorability such identification may not be trivial. Existing methods used to address this problem often require reliance on parametric assumptions and most, if not all, ignore the information embedded in the response variable.We distinguish between the concepts of "common support" and "common causal support." We propose a new approach for identifying common causal support that addresses some of the shortcomings of existing methods. We motivate and illustrate the approach using data from the National Longitudinal Survey of Youth to estimate the effect of breastfeeding at least nine months on reading and math achievement scores at age five or six. We also evaluate the comparative performance of this method in hypothetical examples and simulations where the true treatment effect is known.

Original languageEnglish (US)
Pages (from-to)1386-1420
Number of pages35
JournalAnnals of Applied Statistics
Volume7
Issue number3
DOIs
StatePublished - Sep 2013

Fingerprint

Causal Inference
Bayesian Nonparametrics
Ignorability
Unit
Covariates
Causal Effect
Observational Study
Treatment Effects
Estimate
Trivial
Extremes
High-dimensional
Sufficient
Evaluate
Children
Causal inference
Breast feeding
Simulation
Inference

Keywords

  • BART
  • Breastfeeding
  • Common support
  • Overlap
  • Propensity scores

ASJC Scopus subject areas

  • Statistics, Probability and Uncertainty
  • Modeling and Simulation
  • Statistics and Probability

Cite this

@article{94a6da05562144c38e9a7caf3360e422,
title = "Assessing lack of common support in causal inference using bayesian nonparametrics: Implications for evaluating the effect of breastfeeding on children's cognitive outcomes",
abstract = "Causal inference in observational studies typically requires making comparisons between groups that are dissimilar. For instance, researchers investigating the role of a prolonged duration of breastfeeding on child outcomes may be forced to make comparisons between women with substantially different characteristics on average. In the extreme there may exist neighborhoods of the covariate space where there are not sufficient numbers of both groups of women (those who breastfed for prolonged periods and those who did not) to make inferences about those women. This is referred to as lack of common support. Problems can arise when we try to estimate causal effects for units that lack common support, thus we may want to avoid inference for such units. If ignorability is satisfied with respect to a set of potential confounders, then identifying whether, or for which units, the common support assumption holds is an empirical question. However, in the high-dimensional covariate space often required to satisfy ignorability such identification may not be trivial. Existing methods used to address this problem often require reliance on parametric assumptions and most, if not all, ignore the information embedded in the response variable.We distinguish between the concepts of {"}common support{"} and {"}common causal support.{"} We propose a new approach for identifying common causal support that addresses some of the shortcomings of existing methods. We motivate and illustrate the approach using data from the National Longitudinal Survey of Youth to estimate the effect of breastfeeding at least nine months on reading and math achievement scores at age five or six. We also evaluate the comparative performance of this method in hypothetical examples and simulations where the true treatment effect is known.",
keywords = "BART, Breastfeeding, Common support, Overlap, Propensity scores",
author = "Jennifer Hill and Su, {Yu Sung}",
year = "2013",
month = "9",
doi = "10.1214/13-AOAS630",
language = "English (US)",
volume = "7",
pages = "1386--1420",
journal = "Annals of Applied Statistics",
issn = "1932-6157",
publisher = "Institute of Mathematical Statistics",
number = "3",

}

TY - JOUR

T1 - Assessing lack of common support in causal inference using bayesian nonparametrics

T2 - Implications for evaluating the effect of breastfeeding on children's cognitive outcomes

AU - Hill, Jennifer

AU - Su, Yu Sung

PY - 2013/9

Y1 - 2013/9

N2 - Causal inference in observational studies typically requires making comparisons between groups that are dissimilar. For instance, researchers investigating the role of a prolonged duration of breastfeeding on child outcomes may be forced to make comparisons between women with substantially different characteristics on average. In the extreme there may exist neighborhoods of the covariate space where there are not sufficient numbers of both groups of women (those who breastfed for prolonged periods and those who did not) to make inferences about those women. This is referred to as lack of common support. Problems can arise when we try to estimate causal effects for units that lack common support, thus we may want to avoid inference for such units. If ignorability is satisfied with respect to a set of potential confounders, then identifying whether, or for which units, the common support assumption holds is an empirical question. However, in the high-dimensional covariate space often required to satisfy ignorability such identification may not be trivial. Existing methods used to address this problem often require reliance on parametric assumptions and most, if not all, ignore the information embedded in the response variable.We distinguish between the concepts of "common support" and "common causal support." We propose a new approach for identifying common causal support that addresses some of the shortcomings of existing methods. We motivate and illustrate the approach using data from the National Longitudinal Survey of Youth to estimate the effect of breastfeeding at least nine months on reading and math achievement scores at age five or six. We also evaluate the comparative performance of this method in hypothetical examples and simulations where the true treatment effect is known.

AB - Causal inference in observational studies typically requires making comparisons between groups that are dissimilar. For instance, researchers investigating the role of a prolonged duration of breastfeeding on child outcomes may be forced to make comparisons between women with substantially different characteristics on average. In the extreme there may exist neighborhoods of the covariate space where there are not sufficient numbers of both groups of women (those who breastfed for prolonged periods and those who did not) to make inferences about those women. This is referred to as lack of common support. Problems can arise when we try to estimate causal effects for units that lack common support, thus we may want to avoid inference for such units. If ignorability is satisfied with respect to a set of potential confounders, then identifying whether, or for which units, the common support assumption holds is an empirical question. However, in the high-dimensional covariate space often required to satisfy ignorability such identification may not be trivial. Existing methods used to address this problem often require reliance on parametric assumptions and most, if not all, ignore the information embedded in the response variable.We distinguish between the concepts of "common support" and "common causal support." We propose a new approach for identifying common causal support that addresses some of the shortcomings of existing methods. We motivate and illustrate the approach using data from the National Longitudinal Survey of Youth to estimate the effect of breastfeeding at least nine months on reading and math achievement scores at age five or six. We also evaluate the comparative performance of this method in hypothetical examples and simulations where the true treatment effect is known.

KW - BART

KW - Breastfeeding

KW - Common support

KW - Overlap

KW - Propensity scores

UR - http://www.scopus.com/inward/record.url?scp=84885066656&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84885066656&partnerID=8YFLogxK

U2 - 10.1214/13-AOAS630

DO - 10.1214/13-AOAS630

M3 - Article

AN - SCOPUS:84885066656

VL - 7

SP - 1386

EP - 1420

JO - Annals of Applied Statistics

JF - Annals of Applied Statistics

SN - 1932-6157

IS - 3

ER -