Separators and Adjustment Sets in Causal Graphs: Complete Criteria and an Algorithmic Framework

Abstract Principled reasoning about the identifiability of causal effects from non-experimental data is an important application of graphical causal models. This paper focuses on effects that are identifiable by covariate adjustment, a commonly used estimation approach. We present an algorithmic framework for efficiently testing, constructing, and enumerating m-separators in ancestral graphs (AGs), a class of graphical causal models that can represent uncertainty about the presence of latent confounders. Furthermore, we prove a reduction from causal effect identification by covariate adjustment to m-separation in a subgraph for directed acyclic graphs (DAGs) and maximal ancestral graphs (MAGs). Jointly, these results yield constructive criteria that characterize all adjustment sets as well as all minimal and minimum adjustment sets for identification of a desired causal effect with multiple exposures and outcomes in the presence of latent confounding. Our results extend several existing solutions for special cases of these problems. Our efficient algorithms allowed us to empirically quantify the identifiability gap between covariate adjustment and the do-calculus in random DAGs and MAGs, covering a wide range of scenarios. Implementations of our algorithms are provided in the R package dagitty .

[1]  P. Spirtes,et al.  Ancestral graph Markov models , 2002 .

[2]  Marloes H. Maathuis,et al.  Interpreting and Using CPDAGs With Background Knowledge , 2017, UAI.

[3]  A. Dawid Conditional Independence in Statistical Theory , 1979 .

[4]  Edward Ott,et al.  Growing networks with geographical attachment preference: emergence of small worlds. , 2004, Physical review. E, Statistical, nonlinear, and soft matter physics.

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

[6]  Luis M. de Campos,et al.  An Algorithm for Finding Minimum d-Separating Sets in Belief Networks , 1996, UAI.

[7]  Judea Pearl,et al.  Testing regression models with fewer regressors , 2011, AISTATS.

[8]  Peter Bühlmann,et al.  Causal Inference Using Graphical Models with the R Package pcalg , 2012 .

[9]  Elias Bareinboim,et al.  Causal Effect Identification by Adjustment under Confounding and Selection Biases , 2017, AAAI.

[10]  Johannes Textor,et al.  Complete Graphical Characterization and Construction of Adjustment Sets in Markov Equivalence Classes of Ancestral Graphs , 2016, J. Mach. Learn. Res..

[11]  Adam N. Glynn,et al.  Front-Door Versus Back-Door Adjustment With Unmeasured Confounding: Bias Formulas for Front-Door and Hybrid Adjustments With Application to a Job Training Program , 2018, Journal of the American Statistical Association.

[12]  M. Graffar [Modern epidemiology]. , 1971, Bruxelles medical.

[13]  David S. Johnson,et al.  Computers and Intractability: A Guide to the Theory of NP-Completeness , 1978 .

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

[15]  Marco Valtorta,et al.  Pearl's Calculus of Intervention Is Complete , 2006, UAI.

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

[17]  James B. Orlin,et al.  Max flows in O(nm) time, or better , 2013, STOC '13.

[18]  Judea Pearl,et al.  Identification of Conditional Interventional Distributions , 2006, UAI.

[19]  Duncan J. Watts,et al.  Collective dynamics of ‘small-world’ networks , 1998, Nature.

[20]  P. Spirtes,et al.  Causation, Prediction, and Search, 2nd Edition , 2001 .

[21]  Albert,et al.  Emergence of scaling in random networks , 1999, Science.

[22]  Luis M. de Campos,et al.  Searching for Bayesian Network Structures in the Space of Restricted Acyclic Partially Directed Graphs , 2011, J. Artif. Intell. Res..

[23]  Shimon Even,et al.  Graph Algorithms , 1979 .

[24]  Judea Pearl,et al.  Identification of Joint Interventional Distributions in Recursive Semi-Markovian Causal Models , 2006, AAAI.

[25]  Illtyd Trethowan Causality , 1938 .

[26]  Maciej Liskiewicz,et al.  Separators and Adjustment Sets in Markov Equivalent DAGs , 2016, AAAI.

[27]  S Greenland,et al.  Hierarchical regression for epidemiologic analyses of multiple exposures. , 1994, Environmental health perspectives.

[28]  Maciej Liskiewicz,et al.  On Searching for Generalized Instrumental Variables , 2016, AISTATS.

[29]  Diego Colombo,et al.  A generalized backdoor criterion , 2013, ArXiv.

[30]  Johannes Textor,et al.  A Complete Generalized Adjustment Criterion , 2015, UAI.

[31]  Amitabh Sagar,et al.  Author's reply , 1991, Journal of neurosciences in rural practice.

[32]  Maciej Liskiewicz,et al.  Efficiently Finding Conditional Instruments for Causal Inference , 2015, IJCAI.

[33]  Santtu Tikka,et al.  Identifying Causal Effects with the R Package causaleffect , 2017, 1806.07161.

[34]  Felix Elwert,et al.  Graphical Causal Models , 2013 .

[35]  Maciej Liskiewicz,et al.  Robust causal inference using Directed Acyclic Graphs: the R package , 2018 .

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

[37]  Maciej Liskiewicz,et al.  Adjustment Criteria in Causal Diagrams: An Algorithmic Perspective , 2011, UAI.

[38]  P. Spirtes,et al.  Causation, prediction, and search , 1993 .

[39]  S. Greenland,et al.  Limitations of individual causal models, causal graphs, and ignorability assumptions, as illustrated by random confounding and design unfaithfulness , 2015, European Journal of Epidemiology.

[40]  Ian Shrier,et al.  Re: The design versus the analysis of observational studies for causal effects: parallels with the design of randomized trials. , 2008, Statistics in medicine.

[41]  Y. Rosseel,et al.  Local fit evaluation of structural equation models using graphical criteria. , 2017, Psychological methods.

[42]  Eric J. Tchetgen Tchetgen,et al.  Robust inference on population indirect causal effects: the generalized front door criterion , 2017, Journal of the Royal Statistical Society. Series B, Statistical methodology.

[43]  Jiji Zhang,et al.  Causal Reasoning with Ancestral Graphs , 2008, J. Mach. Learn. Res..

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

[45]  S. Morgan Handbook of Causal Analysis for Social Research , 2013 .

[46]  Jin Tian,et al.  Finding Minimal D-separators , 1998 .

[47]  T. VanderWeele On the Relative Nature of Overadjustment and Unnecessary Adjustment , 2009, Epidemiology.

[48]  G. Sivyer,et al.  Letter to the Editor , 2014, Dermatology practical & conceptual.