Causal Inference in Statistics: A Primer

By Judea Pearl, M. Glymour, Nicholas P. Jewell. 2016

1,632 citations recorded by the source corpus.

How researchers see this paper

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

In Causal Inference & Discovery

Citations in the corpus

Counterfactual Fairness (2017, background): “1 of [25] and the Supplementary Material2.”
A Causal Bayesian Networks Viewpoint on Fairness (2018, neutral): “The back-door criterion can also be derived from the rules of do-calculus [39,40], which indicate whether and how p→A(Y |A) can be estimated using observations from G: for many graph structures with unobserved confounders the only way to compute causal effects is by collecting observations directly ”
Path-Specific Counterfactual Fairness (2018, methodology): “The suggestion is to view unfairness as the presence of an unfair causal effect of the sensitive attribute on the decision, as already done in Pearl (2000) and Pearl et al. (2016), e.g. , to analyse the case of Berkeley’s alleged sex bias in graduate admission.”
Causally Correct Partial Models for Reinforcement Learning (2020, background): “…changes the original graph from Interventions: The operation of setting x to a ﬁxed value x 0 independently of its parents, known as the do-operator (Pearl et al., 2016), changes the data-generation process to p ( u ) δ ( x − x 0 ) p ( y | x, u ) , where δ ( x − x 0 ) is the delta-function.”
Causal Reasoning from Meta-reinforcement Learning (2019, methodology): “| The rules of do-calculus (Pearl, 2000; Pearl et al., 2016) tell us how to compute p → E = e ( H | E = e ) using observations from G .”
Distilling Causal Effect of Data in Class-Incremental Learning (2021, methodology): “We call this Colliding Effect [31], which can be intuitively understood through a real-life example: a causal graph Intelligence → Grades ← Effort representing a student’s grades are determined by both the intelligence and hard work.”
Interventional Few-Shot Learning (2020, methodology): “• We begin with a Structural Causal Model (SCM) assumption in Section 2.2, which shows that the pre-trained knowledge is essentially a confounder that causes spurious correlations between the sample features and class labels in support set.”
Unbiased Scene Graph Generation From Biased Training (2020, positive): “The above operation is termed intervention in causal inference [40].”
A survey of algorithmic recourse: definitions, formulations, solutions, and prospects (2020, methodology): “For XAI, see [1, 2, 31, 55, 61, 67, 75, 111, 120, 128], for causal and counterfactual inference, see [63, 76, 131, 145, 146, 159, 185], and for optimization, see [32, 137, 166].”
Causal Analysis of Agent Behavior for AI Safety (2021, background): “We refer the reader to Pearl et al. (2016) for an introduction to causality and Pearl (2009) for a comprehensive treatment.”
Causal and Counterfactual Inference (2018, methodology): “For example, the effect of additive interventions, such as “add 5 mg/l of insulin to a given patient” can be estimated using the do-calculus (Pearl et al., 2016, p. 109).”
Visual Commonsense R-CNN (2020, background): “Due to the fact that the causality from the confounders as the category averaged features are not yet veriﬁed, that is, Z may contain colliders (or v-structure) [49] causing spurious correlations when intervention.”
Two Causal Principles for Improving Visual Dialog (2019, methodology): “In particular, we use the Pearl’s structural causal model (SCM) proposed by [28] to hypothesize the data generation process, which is a model-agnostic framework that reﬂects the nature of the data.”
Causal inference and the data-fusion problem (2016, background): “There are different designs (Bottom) showing that data come from nonidealized conditions, specifically: (1) from the same population under an observational regime, PðvÞ; (2) from the same population under an experimental regime when Z is randomized, PðvjdoðzÞÞ; (3) from the same population under sam”
On conditional parity as a notion of non-discrimination in machine learning (2017, methodology): “We wrap up our brief overview of counterfactuals with an example and refer to Pearl, Glymour and Jewell (2016), Chapter 4 for further details.”
Decoupled Classifiers for Group-Fair and Efficient Machine Learning (2017, background): “See the beautiful primer of Pearl et al. (2016) for an introduction to the central concepts and machinery.”
Decoupled classifiers for fair and efficient machine learning (2017, background): “See the beautiful primer of Pearl et al. (2016) for an introduction to the central concepts and machinery.”
Robust causal inference using directed acyclic graphs: the R package 'dagitty'. (2017, background): “For those studying causal inference, there is also an R vignette (available at http://dagitty.net/primer/ ), which shows how to solve many of the exercises in the recent textbook on causality by Pearl et al. (17).”