Counterfactual Graphical Models for Longitudinal Mediation Analysis With Unobserved Confounding

By Ilya Shpitser. 2012

119 citations recorded by the source corpus.

Questions concerning mediated causal effects are of great interest in psychology, cognitive science, medicine, social science, public health, and many other disciplines. For instance, about 60% of recent papers published in leading journals in social psychology contain at least one mediation test (Rucker, Preacher, Tormala, & Petty, 2011). Standard parametric approaches to mediation analysis employ regression models, and either the "difference method" (Judd & Kenny, 1981), more common in epidemiology, or the "product method" (Baron & Kenny, 1986), more common in the social sciences. In this article, we first discuss a known, but perhaps often unappreciated, fact that these parametric approaches are a special case of a general counterfactual framework for reasoning about causality first described by Neyman (1923) and Rubin (1924) and linked to causal graphical models by Robins (1986) and Pearl (2006). We then show a number of advantages of this framework. First, it makes the strong assumptions underlying mediation analysis explicit. Second, it avoids a number of problems present in the product and difference methods, such as biased estimates of effects in certain cases. Finally, we show the generality of this framework by proving a novel result which allows mediation analysis to be applied to longitudinal settings with unobserved confounders.

Subfields: Experimental and Cognitive Psychology, Management Science and Operations Research, Statistics and Probability, Probabilistic & Bayesian Methods, Causal Inference & Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Experimental and Cognitive Psychology

In Management Science and Operations Research

In Statistics and Probability

In Probabilistic & Bayesian Methods

In Causal Inference & Discovery

Citations in the corpus

Learning Optimal Fair Policies (2018, methodology): “Given an approximation of p∗ learned in this way, Nabi & Shpitser (2018) transform regression problems originally defined on p into regression problems on p∗. In other words, instead of learning a regression function E[Y |M,A, S,X] on the observed data distribution p(Z), they approximate the fair di”
Causal Reasoning for Algorithmic Fairness (2018, methodology): “Assuming that a unit would be set to a factual baseline value a for A , the “unfair propagation” of a counterfactual value a (cid:48) of A could be understood as passing it only through A → X 1 → ˆ Y and A → ˆ Y in Figure 2(b), leaving the inputs “through the other edges” at the baseline [30, 36].”
Path-Specific Counterfactual Fairness (2018, methodology): “In the Appendix, we describe the graphical method introduced in Shpitser (2013) for understanding whether a PSE is identiﬁable.”
A Causal Framework for Discovering and Removing Direct and Indirect Discrimination (2016, methodology): “If the recanting witness criterion is not satisﬁed, the π - speciﬁc e ﬀ ect S E π ( x 2 , x 1 ) can be computed as follows based on (Shpitser 2013).”
On Partial Identification of the Pure Direct Effect (2015, background): “…Albert and Nelson, 2011; Tchetgen Tchetgen, 2011; VanderWeele, 2011; Albert, 2012; Tchetgen Tchetgen and Shpitser, 2012; Wang and Albert, 2012; Shpitser, 2013; Tchetgen Tchetgen, 2013; Tchetgen Tchetgen and Shpitser, 2014; Wang et al., 2013; Albert and Wang, 2015; Hsu et al., 2015), not only…”
Estimation of a Semiparametric Natural Direct Effect Model Incorporating Baseline Covariates (2014, unknown): “generalizes Theorem 4 of Tchetgen Tchetgen & Shpitser (2012). Its proof is given in the Supplementary Material.”
Quantifying an Adherence Path-Specific Effect of Antiretroviral Therapy in the Nigeria PEPFAR Program (2014, background): “Effects along such arbitrary (bundles of) causal pathways are known as path-specific effects (Pearl, 2000; Avin et al., 2005; Shpitser, 2013) and it is possible to define them inductively, which results in a quantity that is a function of a nested counterfactual (Shpitser, 2013).”
Interpretation and identification of causal mediation. (2014, unknown): “…causal notion of “set M to a constant,” and the two coincide only under the “no omitted variables” assumption.12 (See Bollen and Pearl (2013), for further discussion of this important observation which is often overlooked in the potential-outcome literature (e.g., Rubin, 2010; Sobel, 2008).)”
The Sensitivity of Counterfactual Fairness to Unmeasured Confounding (2019, background): “4 See Silva et al. (2007) and Barber (2009) for examples of applications of this idea.”
Causal Analysis of Agent Behavior for AI Safety (2021, background): “To estimate the effect of a given pathway, one can chain together the effects of the individual mechanisms along the path (Shpitser, 2013; Chiappa, 2019).”
CAUSAL INFERENCE WITH A GRAPHICAL HIERARCHY OF INTERVENTIONS (2014, background): “The synthesis of potential outcomes and graphs has been instrumental in much of the recent work on identification of various types of causal parameters such as total effects [14, 25–27, 33], and mediated effects [1, 10, 24].”
Causal and Counterfactual Inference (2018, background): “Shpitser (2013) further provides complete algorithms for identifying natural direct and indirect effects and extends these results to path-specific effects with multiple treatments and multiple outcomes.”