Causal Effect Identification by Adjustment under Confounding and Selection Biases

Controlling for selection and confounding biases are two of the most challenging problems in the empirical sciences as well as in artificial intelligence tasks. Covariate adjustment (or, Backdoor Adjustment) is the most pervasive technique used for controlling confounding bias, but the same is oblivious to issues of sampling selection. In this paper, we introduce a generalized version of covariate adjustment that simultaneously controls for both confounding and selection biases. We first derive a sufficient and necessary condition for recovering causal effects using covariate adjustment from an observational distribution collected under preferential selection. We then relax this setting to consider cases when additional, unbiased measurements over a set of covariates are available for use (e.g., the age and gender distribution obtained from census data). Finally, we present a complete algorithm with polynomial delay to find all sets of admissible covariates for adjustment when confounding and selection biases are simultaneously present and unbiased data is available.

[1]  G. G. Stokes.  "J." , 1890, The New Yale Book of Quotations.

[2]  Jin Tian,et al.  Recovering Causal Effects from Selection Bias , 2015, AAAI.

[3]  N. Jewell,et al.  Some surprising results about covariate adjustment in logistic regression models , 1991 .

[4]  Manabu Kuroki,et al.  On recovering a population covariance matrix in the presence of selection bias , 2006 .

[5]  Joseph Y. Halpern,et al.  Proceedings of the Twenty-Eighth AAAI Conference on Artificial Intelligence , 2014, AAAI 2014.

[6]  Maciej Liskiewicz,et al.  Constructing Separators and Adjustment Sets in Ancestral Graphs , 2014, CI@UAI.

[7]  Bianca Zadrozny,et al.  Learning and evaluating classifiers under sample selection bias , 2004, ICML.

[8]  Jin Tian,et al.  Recovering from Selection Bias in Causal and Statistical Inference , 2014, AAAI.

[9]  Mehryar Mohri,et al.  Sample Selection Bias Correction Theory , 2008, ALT.

[10]  Judea Pearl,et al.  Aspects of Graphical Models Connected with Causality , 2011 .

[11]  Elias Bareinboim,et al.  Controlling Selection Bias in Causal Inference , 2011, AISTATS.

[12]  J. Pearl.  Causality: Models, Reasoning and Inference , 2000 .

[13]  J. Angrist.  Conditional Independence in Sample Selection Models , 1997 .

[14]  J. Robins.  Data, Design, and Background Knowledge in Etiologic Inference , 2001, Epidemiology.

[15]  Niels Keiding,et al.  Graphical models for inference under outcome-dependent sampling , 2010, 1101.0901.

[16]  D. Rubin.  Estimating causal effects of treatments in randomized and nonrandomized studies. , 1974 .

[17]  J. Heckman.  Sample selection bias as a specification error , 1979 .

[18]  J. Pearl.  Causal diagrams for empirical research , 1995 .

[19]  Ken Takata,et al.  Space-optimal, backtracking algorithms to list the minimal vertex separators of a graph , 2010, Discret. Appl. Math..

[20]  M. Pirinen,et al.  Including known covariates can reduce power to detect genetic effects in case-control studies , 2012, Nature Genetics.

[21]  J. Mefford,et al.  The Covariate's Dilemma , 2012, PLoS genetics.

[22]  Elias Bareinboim,et al.  Causal inference and the data-fusion problem , 2016, Proceedings of the National Academy of Sciences.

[23]  James M. Robins,et al.  On the Validity of Covariate Adjustment for Estimating Causal Effects , 2010, UAI.

[24]  Jiji Zhang,et al.  On the completeness of orientation rules for causal discovery in the presence of latent confounders and selection bias , 2008, Artif. Intell..

[25]  Nicole A. Lazar,et al.  Statistical Analysis With Missing Data , 2003, Technometrics.

[26]  Vanessa Didelez,et al.  Recovering from Selection Bias using Marginal Structure in Discrete Models , 2015, ACI@UAI.