Regularization and Confounding in Linear Regression for Treatment Effect Estimation

被引：46

作者：

Hahn, P. Richard ^{[1
]}

Carvalho, Carlos M. ^{[2
]}

Puelz, David ^{[2
]}

He, Jingyu ^{[1
]}

机构：

[1] Univ Chicago, Booth Sch Business, 5807 South Woodlawn Ave, Chicago, IL 60637 USA

[2] Univ Texas Austin, McCombs Sch Business, 2110 Speedway Stop, Austin, TX 78712 USA

来源：

BAYESIAN ANALYSIS | 2018年 / 13卷 / 01期

关键词：

causal inference; observational data; shrinkage estimation; PROPENSITY SCORE; VARIABLE SELECTION; UNCERTAINTY; OUTCOMES;

D O I：

10.1214/16-BA1044

中图分类号：

O1 [数学];

学科分类号：

0701 ; 070101 ;

摘要：

This paper investigates the use of regularization priors in the context of treatment effect estimation using observational data where the number of control variables is large relative to the number of observations. First, the phenomenon of "regularization-induced confounding" is introduced, which refers to the tendency of regularization priors to adversely bias treatment effect estimates by over-shrinking control variable regression coefficients. Then, a simultaneous regression model is presented which permits regularization priors to be specified in a way that avoids this unintentional "re-confounding". The new model is illustrated on synthetic and empirical data.

引用

页码：163 / 182

页数：20

共 27 条

[1] BAYESIAN PROPENSITY SCORE ESTIMATORS: INCORPORATING UNCERTAINTIES IN PROPENSITY SCORES INTO CAUSAL INFERENCE [J].

An, Weihua .

SOCIOLOGICAL METHODOLOGY, VOL 40, 2010, 40 :151-189

[2] Inference on Treatment Effects after Selection among High-Dimensional ControlsaEuro [J].

Belloni, Alexandre ;

Chernozhukov, Victor ;

Hansen, Christian .

REVIEW OF ECONOMIC STUDIES, 2014, 81 (02) :608-650

[3] The horseshoe estimator for sparse signals [J].

Carvalho, Carlos M. ;

Polson, Nicholas G. ;

Scott, James G. .

BIOMETRIKA, 2010, 97 (02) :465-480

[4] The impact of legalized abortion on crime [J].

Donohue, JJ ;

Levitt, SD .

QUARTERLY JOURNAL OF ECONOMICS, 2001, 116 (02) :379-420

[5]

Ertefaie A., 2015, ARXIV151108501, P164

[6] Penalized regression procedures for variable selection in the potential outcomes framework [J].

Ghosh, Debashis ;

Zhu, Yeying ;

Coffman, Donna L. .

STATISTICS IN MEDICINE, 2015, 34 (10) :1645-1658

[7]

Hahn P. R., 2016, ELLIPTICAL SLICE SAM, P168

[8]

Hahn P. R., 2016, BAYESIAN ANAL, DOI [10.1214/16-BA1044SUPP, DOI 10.1214/16-BA1044SUPP]

[9] TREATMENT EFFECTS: A BAYESIAN PERSPECTIVE [J].

Heckman, James J. ;

Lopes, Hedibert F. ;

Piatek, Remi .

ECONOMETRIC REVIEWS, 2014, 33 (1-4) :36-67

[10]

Imbens GW, 2015, CAUSAL INFERENCE FOR STATISTICS, SOCIAL, AND BIOMEDICAL SCIENCES: AN INTRODUCTION, P1, DOI 10.1017/CBO9781139025751

← 1 2 3 →