Computation of maximum likelihood estimates in cyclic structural equation models

DOI10.1214/17-AOS1602zbMATH Open1499.62016arXiv1610.03434WikidataQ128584472 ScholiaQ128584472MaRDI QIDQ1731051FDOQ1731051

Authors: Mathias Drton, Christopher Fox, Y. Samuel Wang 

Publication date: 6 March 2019

Published in: The Annals of Statistics (Search for Journal in Brave)

Abstract: Software for computation of maximum likelihood estimates in linear structural equation models typically employs general techniques from non-linear optimization, such as quasi-Newton methods. In practice, careful tuning of initial values is often required to avoid convergence issues. As an alternative approach, we propose a block-coordinate descent method that cycles through the considered variables, updating only the parameters related to a given variable in each step. We show that the resulting block update problems can be solved in closed form even when the structural equation model comprises feedback cycles. Furthermore, we give a characterization of the models for which the block-coordinate descent algorithm is well-defined, meaning that for generic data and starting values all block optimization problems admit a unique solution. For the characterization, we represent each model by its mixed graph (also known as path diagram), which leads to criteria that can be checked in time that is polynomial in the number of considered variables.

Full work available at URL: https://arxiv.org/abs/1610.03434

Recommendations

• Maximum likelihood estimation of nonlinear structural equation models
• Maximum likelihood estimation and model comparison of nonlinear structural equation models with continuous and polytomous variables
• Computing maximum likelihood estimates in recursive linear models with correlated errors
• The maximum likelihood threshold of a path diagram
• An incomplete data approach to the analysis of covariance structures

zbMATH Keywords

maximum likelihood estimationgraphical modelfeedbackcyclic graphlinear structural equation model

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Point estimation (62F10) Estimation in multivariate analysis (62H12) Probabilistic graphical models (62H22)

Cites Work

• Causation, prediction, and search
• Distributional equivalence and structure learning for bow-free acyclic path diagrams
• Subsampling
• Title not available (Why is that?)
• Title not available (Why is that?)
• Likelihood ratio tests and singularities
• Causality. Models, reasoning, and inference
• Half-trek criterion for generic identifiability of linear structural equation models
• Distinctness of the eigenvalues of a quadratic form in a multivariate sample
• Computing maximum likelihood estimates in recursive linear models with correlated errors
• The Method of Path Coefficients
• Probability distributions with summary graph structure
• Lectures on algebraic statistics
• Learning high-dimensional directed acyclic graphs with latent and selection variables
• Estimation of a covariance matrix with zeros
• The Determinant of the Adjacency Matrix of a Graph
• Multimodality of the likelihood in the bivariate seemingly unrelated regressions model
• Principles and practice of structural equation modeling
• A characterization of Markov equivalence for directed cyclic graphs
• Computation of maximum likelihood estimates in cyclic structural equation models
• A Review of Eight Software Packages for Structural Equation Modeling

Cited In (1)

• Computation of maximum likelihood estimates in cyclic structural equation models

Uses Software

• R
• sem
• Stata
• lavaan
• TETRAD

This page was built for publication: Computation of maximum likelihood estimates in cyclic structural equation models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1731051)