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

Evaluating and rejecting true mediation models: a cautionary note.

Jerald R Herting1

  • 1Psychosocial and Community Health and Department of Sociology, University of Washington, Box 358732, Seattle, Washington 98195 , USA. herting@u.washington.edu

Prevention Science : the Official Journal of the Society for Prevention Research
|December 3, 2002
PubMed
Summary

A common method for testing mediation fails when unobserved errors correlate between mediator and outcome variables. This research highlights a flaw in mediation analysis, particularly with the Baron and Kenny procedures.

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

Socioeconomic factors and mental health among young Asian adults in the United States.

Ethnicity & health·2025
Same author

Integrating Worker Health Education in Community Agencies to Address Immigrant Worker Health: A Pilot Implementation Trial.

Journal of occupational and environmental medicine·2024
Same author

REDESIGNING JUVENILE PROBATION TO ALIGN WITH BEHAVIORAL HEALTH AND POSITIVE DEVELOPMENT PRINCIPLEs: A Quasi-Experimental study.

Criminal justice and behavior·2023
Same author

Supportive relationships in childhood: Does it have a long Reach into health and depression outcomes for immigrants from Latin America?

SSM - population health·2023
Same author

Job satisfaction and job security as moderators in the relationships among job demands, musculoskeletal symptoms, and work performance.

International archives of occupational and environmental health·2023
Same author

Advancing health equity: Organizational characteristics emphasized by health department nurse lead executives.

Public health nursing (Boston, Mass.)·2022

Area of Science:

  • Social Sciences
  • Psychology
  • Statistics

Background:

  • Mediation analysis is crucial for understanding causal pathways in complex models.
  • Traditional mediation assessment often relies on simplified 3-variable models.
  • The Baron and Kenny procedures are widely used for testing mediation.

Purpose of the Study:

  • To identify an overlooked problem in mediation analysis.
  • To demonstrate how a 4-variable model reveals limitations in existing mediation assessment rules.
  • To address the challenge of distinguishing true mediation from spurious direct effects.

Main Methods:

  • The study introduces a 4-variable model to reveal limitations in 3-variable mediation assessments.
  • It examines the impact of unobserved relations (correlated error) between mediator and outcome variables.

Related Experiment Videos

  • The research analyzes the performance of established mediation testing rules under these conditions.

    • Main Results:

    • A 3-variable model can incorrectly suggest direct effects when pure mediation exists, due to correlated error.
    • Established mediation rules, like those by Baron and Kenny, fail to distinguish true mediation from models with spurious direct effects.
    • The presence of unobserved correlated error between mediator and outcome variables fundamentally challenges standard mediation assessment.

    Conclusions:

    • The typical assessment of mediation is fundamentally flawed when unobserved correlated errors are present.
    • A 4-variable approach is necessary to accurately identify mediation.
    • Revisiting and refining mediation analysis methods is essential for reliable causal inference.