On the Testability of Causal Models with Latent and Instrumental Variables

Abstract

Certain causal models involving unmeasured variables induce no independence constraints among the observed variables but imply, nevertheless, inequality contraints on the observed distribution. This paper derives a general formula for such instrumental variables, that is, exogenous variables that directly affect some variables but not all. With the help of this formula, it is possible to test whether a model involving instrumental variables may account for the data, or, conversely, whether a given variables can be deemed instrumental.

Appears in Proceedings of the Eleventh Conference on Uncertainty in Artificial Intelligence (UAI1995)

References in corpus (2)

• Probabilistic Evaluation of Sequential Plans from Causal Models with Hidden Variables
• Testing Identifiability of Causal Effects

Cited by in corpus (12)

• Identifying Independencies in Causal Graphs with Feedback
• Instrumentality Tests Revisited
• Counterfactual Prediction with Deep Instrumental Variables Networks
• Inferring latent structures via information inequalities
• Quantifying the Causal Effects of Conversational Tendencies
• Causal Etiology of the Research of James M. Robins
• Ivy: Instrumental Variable Synthesis for Causal Inference
• Inequality Constraints in Causal Models with Hidden Variables
• Information-theoretic inference of common ancestors
• Semi-Instrumental Variables: A Test for Instrument Admissibility
• Causal reappraisal of the quantum three box paradox
• Instrumental Variable Estimation for Compositional Treatments