Graphical Models for Processing Missing Data

Mohan, Karthika; Pearl, Judea

doi:10.1080/01621459.2021.1874961

Cited by 124 publications

(210 citation statements)

References 50 publications

Supporting

Mentioning

208

Contrasting

Unclassified

Order By: Relevance

“…As we show in the appendix, the SCM would need to encode assumptions regarding the active ingredients as latent causal nodes in a DAG (Mohan & Pearl 2021). Claim 8 demonstrates the inadequacy of the exclusion restriction and SUTVA as a substitute for construct validity, in that each of these is only a special case of assumptions regarding inert ingredients, for example, that θ α characterizes the assignment process or non-causal components of the intervention, without specifying α.…”

Section: Causal Specification For Construct Validitymentioning

confidence: 99%

The Necessity of Construct and External Validity for Deductive Causal Inference

Esterling¹,

Brady²,

Schwitzgebel³

2021

Preprint

View full text Add to dashboard Cite

The credibility revolution has facilitated tremendous progress in the social sciences by advancing design-based strategies that rely on internal validity to deductively identify causal effects. We demonstrate that prioritizing internal validity while neglecting construct and external validity prevents causal generalization and misleadingly converts a deductive claim of causality into a claim based on speculation and exploration -- undermining the very goals of the credibility revolution. We develop a formal framework of causal specification to demonstrate that internal, external and construct validity are jointly necessary for generalized claims regarding a causal effect. If one lacks construct validity, one cannot assign meaningful labels to the cause or to the outcome. If one lacks external validity, one cannot make statements about the conditions required for the cause to occur. Re-balancing considerations of internal, construct and external via causal specification preserves and advances the intent of the credibility revolution to understand causal effects.

show abstract

Section: Causal Specification For Construct Validitymentioning

confidence: 99%

The Necessity of Construct and External Validity for Deductive Causal Inference

Esterling¹,

Brady²,

Schwitzgebel³

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…It is, therefore, possible to use the BN as a framework for identifying when causal hypotheses are identifiable in this rather restricted setting. The associated analyses use various graphically stated criteria-such as the front-door and the back-door criteria-see e.g., [11][12][13]. However, unfortunately, the types of missingness that routinely occur in reliability-and, in particular, those associated with the data we collect when performing routine maintenance-are rarely missing across the original random vector associated with the system in this sort of symmetric way.…”

Section: Causal Identifiability On Chain Event Graphs With Informed Missingnessmentioning

confidence: 99%

Causal Algebras on Chain Event Graphs with Informed Missingness for System Failure

Smith

2021

Entropy

View full text Add to dashboard Cite

Graph-based causal inference has recently been successfully applied to explore system reliability and to predict failures in order to improve systems. One popular causal analysis following Pearl and Spirtes et al. to study causal relationships embedded in a system is to use a Bayesian network (BN). However, certain causal constructions that are particularly pertinent to the study of reliability are difficult to express fully through a BN. Our recent work demonstrated the flexibility of using a Chain Event Graph (CEG) instead to capture causal reasoning embedded within engineers’ reports. We demonstrated that an event tree rather than a BN could provide an alternative framework that could capture most of the causal concepts needed within this domain. In particular, a causal calculus for a specific type of intervention, called a remedial intervention, was devised on this tree-like graph. In this paper, we extend the use of this framework to show that not only remedial maintenance interventions but also interventions associated with routine maintenance can be well-defined using this alternative class of graphical model. We also show that the complexity in making inference about the potential relationships between causes and failures in a missing data situation in the domain of system reliability can be elegantly addressed using this new methodology. Causal modelling using a CEG is illustrated through examples drawn from the study of reliability of an energy distribution network.

show abstract

“…This profound observation is supported indeed by several recent explorations. Problems areas such as "meta-analysis" (also known as "data fusion") and "missing data," which were thought to be purely statistical in nature, turned out to be causal problems in disguise (Bareinboim and Pearl, 2016;Mohan and Pearl, 2021).…”

Section: Two Cultures or Collaborative Symbiosis?mentioning

confidence: 99%