Transportability of Causal Effects: Completeness Results

By Elias Bareinboim, Judea Pearl. 2012

141 citations recorded by the source corpus.

The study of transportability aims to identify conditions under which causal information learned from experiments can be reused in a different environment where only passive observations can be collected. The theory introduced in [Pearl and Bareinboim, 2011] (henceforth [PB, 2011]) defines formal conditions for such transfer but falls short of providing an effective procedure for deciding, given assumptions about differences between the source and target domains, whether transportability is feasible. This paper provides such procedure. It establishes a necessary and sufficient condition for deciding when causal effects in the target domain are estimable from both the statistical information available and the causal information transferred from the experiments. The paper further provides a complete algorithm for computing the transport formula, that is, a way of fusing experimental and observational information to synthesize an estimate of the desired causal relation.

Subfields: Electrochemistry

How researchers see this paper

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

In Electrochemistry

Citations in the corpus

Biases in Generative Art: A Causal Look from the Lens of Art History (2020, methodology): “Z in Figure 6 (ii), causal effects are not transportable [Bareinboim and Pearl, 2012].”
A Survey of Learning Causality with Data: Problems and Methods (2018, background): “In [19], the authors further provide a necessary and sufficient condition to decide, given assumptions about differences between the source and target domains, whether transportability is feasible.”
A General Algorithm for Deciding Transportability of Experimental Results (2013, background): “Acknowledgment: A preliminary version of this article was presented at the 26th AAAI Conference, Toronto, CA, July, 2012 [59].”
Transportability from Multiple Environments with Limited Experiments (2013, methodology): “This is possible if we assume that the structural equations share the same set of arguments, though the functional forms of the equations may vary arbitrarily [11].”
Meta-Transportability of Causal Effects: A Formal Approach (2013, background): “In general, however, a more involved mixture of experimental and observational findings would be necessary to obtain a bias-free estimate of the target relation R, a full characterization of which is given in (Bareinboim and Pearl, 2012).”
Causal and Counterfactual Inference (2018, methodology): “While Theorem 3 does not specify the sequence of rules leading to the needed reduction (if such exists), a complete and effective graphical procedure was devised by Bareinboim & Pearl (2012b), which also synthesizes a transport formula whenever possible.”
A Survey of Learning Causality with Data (2018, background): “ions under which causal knowledge learned from experiments can be reused in different domains with observational data only. A formal definition of causal transportability can be referred to [115]. In [17], the authors further provide a necessary and sufficient condition to decide, given assumptions ”
Transportability from Multiple Environments with Limited Experiments: Completeness Results (2014, background): “There exists an extensive literature concerned with the problem of computability of causal relations from a combination of assumptions and data [21, 22, 7, 13].”
Causal Transportability with Limited Experiments (2013, background): “1 Using a graphical representation called selection diagrams to encode commonalities and differences between environments [PB, 2011], a complete graphical and algorithmic characterization was provided in [Bareinboim and Pearl, 2012b] (henceforth, BP), which decides if and how transportability is fea”
Generalizing causal inferences from randomized trials: counterfactual and graphical identification (2019, methodology): “Bareinboim and Pearl have proposed general methods to assess identifiability and, when possible, to identify the post-intervention distribution in a target population by “transporting” information from other sources [47–49].”
The Do-Calculus Revisited (2012, background): “[Bareinboim and Pearl, 2012b] established a complete and eﬀec-tive graphical procedure of conﬁrming transportability which also synthesizes the transport formula whenever possible.”
Preventing Failures Due to Dataset Shift: Learning Predictive Models That Transport (2018, background): “Transportability: Transportability is a framework for the synthesis of experimental and observational data from multiple environments to answer a statistical or causal query in a prespecified target environment (Pearl and Bareinboim, 2011; Bareinboim and Pearl, 2012).”
Causal inference and the data-fusion problem (2016, background): “It can be shown that transportability is feasible if and only if there exists a sequence of rules that transforms the query expression Q=PðyjdoðxÞ, s* Þ into a form where the do-operator is separated from the S-variables (50).”