Regularization and Confounding in Linear Regression for Treatment Effect Estimation

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal

11 Scopus Citations
View graph of relations

Author(s)

  • P. Richard Hahn
  • Carlos M. Carvalho
  • David Puelz
  • Jingyu He

Detail(s)

Original languageEnglish
Pages (from-to)163-182
Journal / PublicationBayesian Analysis
Volume13
Issue number1
Online published11 Jan 2017
Publication statusPublished - 2018
Externally publishedYes

Abstract

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.

Research Area(s)

  • Causal inference, Observational data, Shrinkage estimation

Citation Format(s)

Regularization and Confounding in Linear Regression for Treatment Effect Estimation. / Hahn, P. Richard; Carvalho, Carlos M.; Puelz, David; He, Jingyu.

In: Bayesian Analysis, Vol. 13, No. 1, 2018, p. 163-182.

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal