Regularization and confounding in linear regression for treatment effect estimation. (English) Zbl 06873722

Hahn, P. Richard; Carvalho, Carlos M.; Puelz, David; He, Jingyu

Bayesian Anal. 13, No. 1, 163-182 (2018).

Summary: 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.

MSC:

62J05

Linear regression; mixed models

Keywords:

causal inference; observational data; shrinkage estimation

Software:

BACprior

References:

[1]	An, W. (2010). “Bayesian propensity score estimators: incorporating uncertainties in propensity scores into causal inference.” Sociological Methodology, 40(1): 151-189.
[2]	Belloni, A., Chernozhukov, V., and Hansen, C. (2014). “Inference on treatment effects after selection among high-dimensional controls.” The Review of Economic Studies, 81(2): 608-650. · Zbl 1409.62142
[3]	Carvalho, C. M., Polson, N. G., and Scott, J. G. (2010). “The horseshoe estimator for sparse signals.” Biometrika, 97(2): 465-480. · Zbl 1406.62021
[4]	Donohue III, J. J. and Levitt, S. D. (2001). “The impact of legalized abortion on crime.” Quarterly Journal of Economics, 379-420.
[5]	Ertefaie, A., Asgharian, M., and Stephens, D. (2015). “Variable selection in causal inference using a simultaneous penalization method.” arXiv:1511.08501.
[6]	Ghosh, D., Zhu, Y., and Coffman, D. L. (2015). “Penalized regression procedures for variable selection in the potential outcomes framework.” Statistics in Medicine, 34(10): 1645-1658.
[7]	Hahn, P. R., He, J., and Lopes, H. (2016). “Elliptical slice sampling for Bayesian shrinkage regression with applications to causal inference.” Technical report, University of Chicago Booth School of Business.
[8]	Hahn, P. R., Carvalho, C. M., He, J., and Puelz, D. (2016). “Supplement to “Regularization and confounding in linear regression for treatment effect estimation”.” Bayesian Analysis. · Zbl 06873722
[9]	Heckman, J. J., Lopes, H. F., and Piatek, R. (2014). “Treatment effects: A Bayesian perspective.” Econometric reviews, 33(1-4): 36-67. · Zbl 1491.62218
[10]	Imbens, G. W. and Rubin, D. B. (2015). Causal inference in statistics, social, and biomedical sciences. Cambridge University Press. · Zbl 1355.62002
[11]	Jacobi, L., Wagner, H., and Frühwirth-Schnatter, S. (2016). “Bayesian treatment effects models with variable selection for panel outcomes with an application to earnings effects of maternity leave.” Journal of Econometrics, 193(1): 234-250. · Zbl 1420.62490
[12]	Leamer, E. E. (1978). Specification searches: Ad hoc inference with nonexperimental data, volume 53. John Wiley & Sons Incorporated. · Zbl 0384.62089
[13]	Leamer, E. E. (1983). “Let’s take the con out of econometrics.” The American Economic Review, 73(1): 31-43.
[14]	Lefebvre, G., Atherton, J., and Talbot, D. (2014). “The effect of the prior distribution in the Bayesian Adjustment for Confounding algorithm.” Computational Statistics & Data Analysis, 70: 227-240. · Zbl 1471.62110
[15]	Levitt, S. D. and Dubner, S. J. (2010). Freakonomics, volume 61. Sperling & Kupfer editori.
[16]	Li, M. and Tobias, J. L. (2014). “Bayesian analysis of treatment effect models.” Bayesian Inference in the Social Sciences, 63-90.
[17]	Liang, F., Paulo, R., Molina, G., Clyde, M. A., and Berger, J. O. (2008). “Mixtures of \(g\) priors for Bayesian variable selection.” Journal of the American Statistical Association, 103(481): 410-423. · Zbl 1335.62026
[18]	McCandless, L. C., Gustafson, P., and Austin, P. C. (2009). “Bayesian propensity score analysis for observational data.” Statistics in Medicine, 28(1): 94-112.
[19]	Morgan, S. L. and Winship, C. (2014). Counterfactuals and causal inference. Cambridge University Press.
[20]	Robins, J. M., Mark, S. D., and Newey, W. K. (1992). “Estimating exposure effects by modelling the expectation of exposure conditional on confounders.” Biometrics, 479-495. · Zbl 0768.62099 · doi:10.2307/2532304
[21]	Rosenbaum, P. R. and Rubin, D. B. (1983). “The central role of the propensity score in observational studies for causal effects.” Biometrika, 70(1): 41-55. · Zbl 0522.62091
[22]	Talbot, D., Lefebvre, G., and Atherton, J. (2015). “The Bayesian causal effect estimation algorithm.” Journal of Causal Inference, 3(2): 207-236.
[23]	Wang, C., Dominici, F., Parmigiani, G., and Zigler, C. M. (2015). “Accounting for uncertainty in confounder and effect modifier selection when estimating average causal effects in generalized linear models.” Biometrics, 71(3): 654-665. · Zbl 1419.62468
[24]	Wang, C., Parmigiani, G., and Dominici, F. (2012). “Bayesian effect estimation accounting for adjustment uncertainty.” Biometrics, 68(3): 661-671. · Zbl 1274.62895
[25]	Wilson, A. and Reich, B. J. (2014). “Confounder selection via penalized credible regions.” Biometrics, 70(4): 852-861. · Zbl 1393.62107
[26]	Zellner, A. (1986). “On assessing prior distributions and Bayesian regression analysis with g-prior distributions.” Bayesian Inference and Decision Techniques: Essays in Honor of Bruno De Finetti, 6: 233-243. · Zbl 0655.62071
[27]	Zigler, C. M. and Dominici, F. (2014). “Uncertainty in propensity score estimation: Bayesian methods for variable selection and model-averaged causal effects.