Determinantal generalizations of instrumental variables

Linear structural equation models relate the components of a random vector using linear interdependencies and Gaussian noise. Each such model can be naturally associated with a mixed graph whose vertices correspond to the components of the random vector. The graph contains directed edges that repres...

Full description

Bibliographic Details
Main Authors: Weihs, Luca, Robinson, Bill, Dufresne, Emilie, Kenkel, Jennifer, Kubjas, Kaie, McGee, Reginald L. II, Nguyen, Nhan, Robeva, Elina, Drton, Mathias
Format: Article
Language:English
Published: De Gruyter 2018
Subjects:
Online Access:https://eprints.nottingham.ac.uk/48026/
_version_ 1848797674686382080
author Weihs, Luca
Robinson, Bill
Dufresne, Emilie
Kenkel, Jennifer
Kubjas, Kaie
McGee, Reginald L. II
Nguyen, Nhan
Robeva, Elina
Drton, Mathias
author_facet Weihs, Luca
Robinson, Bill
Dufresne, Emilie
Kenkel, Jennifer
Kubjas, Kaie
McGee, Reginald L. II
Nguyen, Nhan
Robeva, Elina
Drton, Mathias
author_sort Weihs, Luca
building Nottingham Research Data Repository
collection Online Access
description Linear structural equation models relate the components of a random vector using linear interdependencies and Gaussian noise. Each such model can be naturally associated with a mixed graph whose vertices correspond to the components of the random vector. The graph contains directed edges that represent the linear relationships between components, and bidirected edges that encode unobserved confounding. We study the problem of generic identifiability, that is, whether a generic choice of linear and confounding effects can be uniquely recovered from the joint covariance matrix of the observed random vector. An existing combinatorial criterion for establishing generic identifiability is the half-trek criterion (HTC), which uses the existence of trek systems in the mixed graph to iteratively discover generically invertible linear equation systems in polynomial time. By focusing on edges one at a time, we establish new sufficient and necessary conditions for generic identifiability of edge effects extending those of the HTC. In particular, we show how edge coefficients can be recovered as quotients of subdeterminants of the covariance matrix, which constitutes a determinantal generalization of formulas obtained when using instrumental variables for identification.
first_indexed 2025-11-14T20:07:38Z
format Article
id nottingham-48026
institution University of Nottingham Malaysia Campus
institution_category Local University
language English
last_indexed 2025-11-14T20:07:38Z
publishDate 2018
publisher De Gruyter
recordtype eprints
repository_type Digital Repository
spelling nottingham-480262018-12-08T04:30:16Z https://eprints.nottingham.ac.uk/48026/ Determinantal generalizations of instrumental variables Weihs, Luca Robinson, Bill Dufresne, Emilie Kenkel, Jennifer Kubjas, Kaie McGee, Reginald L. II Nguyen, Nhan Robeva, Elina Drton, Mathias Linear structural equation models relate the components of a random vector using linear interdependencies and Gaussian noise. Each such model can be naturally associated with a mixed graph whose vertices correspond to the components of the random vector. The graph contains directed edges that represent the linear relationships between components, and bidirected edges that encode unobserved confounding. We study the problem of generic identifiability, that is, whether a generic choice of linear and confounding effects can be uniquely recovered from the joint covariance matrix of the observed random vector. An existing combinatorial criterion for establishing generic identifiability is the half-trek criterion (HTC), which uses the existence of trek systems in the mixed graph to iteratively discover generically invertible linear equation systems in polynomial time. By focusing on edges one at a time, we establish new sufficient and necessary conditions for generic identifiability of edge effects extending those of the HTC. In particular, we show how edge coefficients can be recovered as quotients of subdeterminants of the covariance matrix, which constitutes a determinantal generalization of formulas obtained when using instrumental variables for identification. De Gruyter 2018-03-26 Article PeerReviewed application/pdf en https://eprints.nottingham.ac.uk/48026/1/1702.03884.pdf Weihs, Luca, Robinson, Bill, Dufresne, Emilie, Kenkel, Jennifer, Kubjas, Kaie, McGee, Reginald L. II, Nguyen, Nhan, Robeva, Elina and Drton, Mathias (2018) Determinantal generalizations of instrumental variables. Journal of Causal Inference, 6 (1). p. 20170009. ISSN 2193-3685 trek separation; half-trek criterion; structural equation models; identifiability generic identifiability https://www.degruyter.com/view/j/jci.ahead-of-print/jci-2017-0009/jci-2017-0009.xml doi:10.1515/jci-2017-0009 doi:10.1515/jci-2017-0009
spellingShingle trek separation; half-trek criterion; structural equation models; identifiability
generic identifiability
Weihs, Luca
Robinson, Bill
Dufresne, Emilie
Kenkel, Jennifer
Kubjas, Kaie
McGee, Reginald L. II
Nguyen, Nhan
Robeva, Elina
Drton, Mathias
Determinantal generalizations of instrumental variables
title Determinantal generalizations of instrumental variables
title_full Determinantal generalizations of instrumental variables
title_fullStr Determinantal generalizations of instrumental variables
title_full_unstemmed Determinantal generalizations of instrumental variables
title_short Determinantal generalizations of instrumental variables
title_sort determinantal generalizations of instrumental variables
topic trek separation; half-trek criterion; structural equation models; identifiability
generic identifiability
url https://eprints.nottingham.ac.uk/48026/
https://eprints.nottingham.ac.uk/48026/
https://eprints.nottingham.ac.uk/48026/