Propensity score-matching methods for nonexperimental causal studies

Rajeev Dehejia, Sadek Wahba

Research output: Contribution to journalReview article

Abstract

This paper considers causal inference and sample selection bias in nonexperimental settings in which (i) few units in the nonexperimental comparison group are comparable to the treatment units, and (ii) selecting a subset of comparison units similar to the treatment units is difficult because units must be compared across a high-dimensional set of pretreatment characteristics. We discuss the use of propensity score-matching methods, and implement them using data from the National Supported Work experiment. Following LaLonde (1986), we pair the experimental treated units with nonexperimental comparison units from the CPS and PSID, and compare the estimates of the treatment effect obtained using our methods to the benchmark results from the experiment. For both comparison groups, we show that the methods succeed in focusing attention on the small subset of the comparison units comparable to the treated units and, hence, in alleviating the bias due to systematic differences between the treated and comparison units.

Original languageEnglish (US)
Pages (from-to)151-161
Number of pages11
JournalReview of Economics and Statistics
Volume84
Issue number1
DOIs
StatePublished - Feb 2002

Fingerprint

experiment
trend
Matching method
Propensity score matching
Group
Experiment
Benchmark
Pretreatment
Treatment effects
Causal inference
Sample selection bias

ASJC Scopus subject areas

  • Economics and Econometrics
  • Social Sciences (miscellaneous)

Cite this

Propensity score-matching methods for nonexperimental causal studies. / Dehejia, Rajeev; Wahba, Sadek.

In: Review of Economics and Statistics, Vol. 84, No. 1, 02.2002, p. 151-161.

Research output: Contribution to journalReview article

@article{151951e549cd4c5fb210e8b23bd4dbd5,
title = "Propensity score-matching methods for nonexperimental causal studies",
abstract = "This paper considers causal inference and sample selection bias in nonexperimental settings in which (i) few units in the nonexperimental comparison group are comparable to the treatment units, and (ii) selecting a subset of comparison units similar to the treatment units is difficult because units must be compared across a high-dimensional set of pretreatment characteristics. We discuss the use of propensity score-matching methods, and implement them using data from the National Supported Work experiment. Following LaLonde (1986), we pair the experimental treated units with nonexperimental comparison units from the CPS and PSID, and compare the estimates of the treatment effect obtained using our methods to the benchmark results from the experiment. For both comparison groups, we show that the methods succeed in focusing attention on the small subset of the comparison units comparable to the treated units and, hence, in alleviating the bias due to systematic differences between the treated and comparison units.",
author = "Rajeev Dehejia and Sadek Wahba",
year = "2002",
month = "2",
doi = "10.1162/003465302317331982",
language = "English (US)",
volume = "84",
pages = "151--161",
journal = "Review of Economics and Statistics",
issn = "0034-6535",
publisher = "MIT Press Journals",
number = "1",

}

TY - JOUR

T1 - Propensity score-matching methods for nonexperimental causal studies

AU - Dehejia, Rajeev

AU - Wahba, Sadek

PY - 2002/2

Y1 - 2002/2

N2 - This paper considers causal inference and sample selection bias in nonexperimental settings in which (i) few units in the nonexperimental comparison group are comparable to the treatment units, and (ii) selecting a subset of comparison units similar to the treatment units is difficult because units must be compared across a high-dimensional set of pretreatment characteristics. We discuss the use of propensity score-matching methods, and implement them using data from the National Supported Work experiment. Following LaLonde (1986), we pair the experimental treated units with nonexperimental comparison units from the CPS and PSID, and compare the estimates of the treatment effect obtained using our methods to the benchmark results from the experiment. For both comparison groups, we show that the methods succeed in focusing attention on the small subset of the comparison units comparable to the treated units and, hence, in alleviating the bias due to systematic differences between the treated and comparison units.

AB - This paper considers causal inference and sample selection bias in nonexperimental settings in which (i) few units in the nonexperimental comparison group are comparable to the treatment units, and (ii) selecting a subset of comparison units similar to the treatment units is difficult because units must be compared across a high-dimensional set of pretreatment characteristics. We discuss the use of propensity score-matching methods, and implement them using data from the National Supported Work experiment. Following LaLonde (1986), we pair the experimental treated units with nonexperimental comparison units from the CPS and PSID, and compare the estimates of the treatment effect obtained using our methods to the benchmark results from the experiment. For both comparison groups, we show that the methods succeed in focusing attention on the small subset of the comparison units comparable to the treated units and, hence, in alleviating the bias due to systematic differences between the treated and comparison units.

UR - http://www.scopus.com/inward/record.url?scp=0036102924&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0036102924&partnerID=8YFLogxK

U2 - 10.1162/003465302317331982

DO - 10.1162/003465302317331982

M3 - Review article

VL - 84

SP - 151

EP - 161

JO - Review of Economics and Statistics

JF - Review of Economics and Statistics

SN - 0034-6535

IS - 1

ER -