Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Cross-validation by downweighting influential cases in structural equation modelling.

Ke-Hai Yuan1, Linda L Marshall, Rebecca Weston

  • 1Dept of Psychology, University of Notre Dame, IN 46556, USA. kyuan@nd.edu

The British Journal of Mathematical and Statistical Psychology
|May 30, 2002
PubMed
Summary

Cross-validation (CV) in structural equation modeling is sensitive to outliers. Using robust covariance matrices improves model selection accuracy by reducing outlier influence, leading to more reliable results in social and behavioral sciences.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Trauma film paradigm in college students: Validation of a virtual methodology.

Psychological trauma : theory, research, practice and policy·2026
Same author

Signal-to-Noise Ratio in Estimating and Testing the Mediation Effect: Structural Equation Modeling versus Path Analysis with Weighted Composites.

Psychometrika·2026
Same author

Servant Leadership in Higher Education: A Graded Response Model Approach to Item Response Theory Analysis.

Psychological reports·2025
Same author

Signal-to-Noise Ratio in Estimating and Testing the Mediation Effect: Structural Equation Modeling versus Path Analysis with Weighted Composites.

Psychometrika·2024
Same author

Which method delivers greater signal-to-noise ratio: Structural equation modelling or regression analysis with weighted composites?

The British journal of mathematical and statistical psychology·2023
Same author

Replies to comments on "Which method delivers greater signal-to-noise ratio: Structural equation modelling or regression analysis with weighted composites?" by Yuan and Fang (2023).

The British journal of mathematical and statistical psychology·2023

Area of Science:

  • Social and behavioral sciences
  • Statistical modeling
  • Psychometrics

Background:

  • Structural equation modeling (SEM) is prevalent for testing theories and causal relationships among latent constructs.
  • Cross-validation (CV) is essential for selecting the optimal model from competing SEM structures.
  • Real-world data frequently contain influential cases or outliers, potentially compromising standard CV procedures.

Purpose of the Study:

  • To address the limitations of traditional cross-validation (CV) methods in structural equation modeling (SEM) when applied to datasets with outliers.
  • To introduce and develop a novel procedure for utilizing robust covariance matrices within the model calibration and validation phases of SEM.
  • To enhance the reliability and validity of model selection in the presence of influential data points.

Related Experiment Videos

Main Methods:

  • The study critically examines the inherent drawbacks of CV techniques that rely on sample covariance matrices.
  • A new procedure is proposed, integrating robust covariance matrices into both the calibration and validation steps of SEM.
  • Empirical examples are used to demonstrate the practical application and effectiveness of the proposed method.

Main Results:

  • CV indices derived from sample covariance matrices exhibit high sensitivity to influential cases, with single outliers capable of misdirecting model selection.
  • The proposed CV index, utilizing robust covariance matrices, demonstrates significantly reduced sensitivity to outliers.
  • This robustness ensures more accurate and dependable model selection, even in the presence of data anomalies.

Conclusions:

  • Standard CV methods in SEM are unreliable when influential cases are present in the data.
  • Employing robust covariance matrices in CV provides a more stable and valid approach to model selection.
  • The developed procedure enhances the practical utility of SEM by yielding more trustworthy conclusions about model structures.