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

GEE analysis of negatively correlated binary responses: a caution.

J A Hanley1, A Negassa, M D Edwardes

  • 1Department of Epidemiology and Biostatistics, McGill University, Montreal, Canada. Jimh@epid.lan.mcgill.ca

Statistics in Medicine
|March 4, 2000
PubMed
Summary

Generalized estimating equations (GEE) can yield unreliable results for correlated binary data with low variation. Researchers recommend using independence working correlations and robust standard errors as a more trustworthy alternative for accurate analysis.

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

Time-dependent prognostic accuracy measures for recurrent event data.

Biometrics·2024
Same author

Inference for covariate-adjusted time-dependent prognostic accuracy measures.

Statistics in medicine·2023
Same author

Screening histories and contact with physicians as determinants of cervical cancer risk in Montreal, Quebec.

Current oncology (Toronto, Ont.)·2014
Same author

Impact of HLA mismatch at first kidney transplant on lifetime with graft function in young recipients.

American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons·2014
Same author

Risk of prostate cancer death in long-term users of warfarin: a population-based case-control study.

Cancer causes & control : CCC·2013
Same author

Change in mortality risk over time in young kidney transplant recipients.

American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons·2011

Area of Science:

  • Statistics
  • Biostatistics
  • Longitudinal Data Analysis

Background:

  • Generalized estimating equations (GEE) are widely used for correlated data.
  • Standard GEE implementations may face challenges with specific data characteristics.

Purpose of the Study:

  • To investigate the reliability of GEE with exchangeable correlation structures for binary responses.
  • To identify alternative methods when standard GEE assumptions are violated.

Main Methods:

  • Analysis of longitudinal and correlated response data using generalized estimating equations.
  • Evaluation of GEE performance under conditions of less than binomial variation.
  • Comparison of exchangeable versus independence working correlation structures.

Related Experiment Videos

Main Results:

  • Prevailing GEE software may produce unreliable results when binary responses exhibit less than binomial variation over clusters.
  • Modeling with exchangeable correlations can be problematic in such scenarios.
  • Bounding negative correlation is not always a sufficient solution.

Conclusions:

  • For binary data with low variation, using independence working correlation and robust standard errors is a more reliable approach.
  • Researchers should exercise caution when applying standard GEE with exchangeable correlations to heterogeneous clustered data.
  • Alternative correlation structures and robust inference methods are crucial for trustworthy GEE analysis.