Measurement Error and Nonlinear Confounding: Structural Equation Modeling Versus Double Machine Learning

Abstract

A latent confounder measured by fallible indicators can be adjusted for with structural equation modeling (SEM), which corrects measurement error but requires a specified functional form, or with double machine learning (DML), which learns the form but ignores measurement error. We derive both population biases and the crossover reliability, the composite reliability above which DML with the indicators or their mean as inputs is less biased than linear SEM; it depends on the nonlinear share of confounding and falls when covariates predict the confounder. A 46-condition Monte Carlo study confirmed both biases and the predicted ordering. With gradient boosting, learning the conditional means raised the DML bias by up to 0.05 and moved the finite-sample crossover upward. A quadratic SEM was unbiased when its form was correct and removed two thirds of the linear-SEM bias under a saturating nonlinearity but little of it under a cubic one.

Dynamic Structural Equation Modeling in Blimp
Brian T Keller
PsyArXiv8 Sep 2026PreprintAll rights reservedFull textStatisticsOSF
P-stationary structural equation modeling: Causal effect identification under temporally incomplete observations
Nicolas-Domenic Reiter
arXiv1 Nov 2026PreprintCC BY 4.0Full textStatisticsarXiv
MASEMiner – AI-Assisted Data Extraction for Meta-Analytic Structural Equation Modeling
Ulrich Schroeders, Timo Gnambs, Johanna Einsiedler
PsyArXiv25 Oct 2026PreprintCC BY 4.0Full textStatisticsOSF
Unraveling the shared genetic architecture of metabolic bone diseases via genomic structural equation modeling and multi-omics integration
Kun Yi, Fan Wang, Feiping Hu +10
PLOS One27 Oct 2026Peer-reviewedCC BY 4.0Full textEnzymology & metabolism Bioinformatics & computational biologyPLOSPMC
Causal Path Analysis from Perturbational and Population-Scale Single-Cell Data with Multiscale Confounding and Measurement Error
Kwangmoon Park, Hongzhe Li
arXiv15 Sep 2026PreprintCC BY-NC-ND 4.0Full textStatisticsarXiv

Cite

BibTeX RIS

BibTeX

@misc{song2026measurement,
  title = {{Measurement Error and Nonlinear Confounding: Structural Equation Modeling Versus Double Machine Learning}},
  author = {Eunjeong Song and Sehee Hong},
  howpublished = {arXiv},
  year = {2026},
  month = nov,
  doi = {10.48550/arxiv.2611.01375},
  eprint = {2611.01375},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.01375},
}

RIS

TY  - UNPB
TI  - Measurement Error and Nonlinear Confounding: Structural Equation Modeling Versus Double Machine Learning
AU  - Eunjeong Song
AU  - Sehee Hong
PY  - 2026
DA  - 2026/11/01
PB  - arXiv
DO  - 10.48550/arxiv.2611.01375
UR  - https://arxiv.org/abs/2611.01375
AB  - A latent confounder measured by fallible indicators can be adjusted for with structural equation modeling (SEM), which corrects measurement error but requires a specified functional form, or with double machine learning (DML), which learns the form but ignores measurement error. We derive both population biases and the crossover reliability, the composite reliability above which DML with the indicators or their mean as inputs is less biased than linear SEM; it depends on the nonlinear share of confounding and falls when covariates predict the confounder. A 46-condition Monte Carlo study confirmed both biases and the predicted ordering. With gradient boosting, learning the conditional means raised the DML bias by up to 0.05 and moved the finite-sample crossover upward. A quadratic SEM was unbiased when its form was correct and removed two thirds of the linear-SEM bias under a saturating nonlinearity but little of it under a cubic one.
ER  -

Abstract

Related

Cite