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

Improved hypothesis testing for coefficients in generalized estimating equations with small samples of clusters.

Daniel F McCaffrey1, Robert M Bell

  • 1The RAND Corporation, Pittsburgh, PA 15213, USA. daniel_mccaffrey@rand.org

Statistics in Medicine
|February 4, 2006
PubMed
Summary

This study introduces bias-reduced linearization (BRL) to improve standard error estimation for generalized estimating equations (GEE). The new method offers more accurate variability estimates and better hypothesis testing for clustered data, especially with small cluster numbers.

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

Practical challenges in mediation analysis: a guide for applied researchers.

Health services & outcomes research methodology·2025
Same author

SGLT1 contributes to glucose-mediated exacerbation of ischemia-reperfusion injury in ex vivo rat heart.

Basic research in cardiology·2024
Same author

Sampling weighting strategies in causal mediation analysis.

BMC medical research methodology·2024
Same author

Estimating generalized propensity scores with survey and attrition weighted data.

Statistics in medicine·2024
Same author

A Tutorial for Propensity Score Weighting for Moderation Analysis With Categorical Variables: An Application Examining Smoking Disparities Among Sexual Minority Adults.

Medical care·2023
Same author

Comparative effectiveness of individual versus family-based substance use treatment on adolescent self-injurious thoughts and behaviors.

Journal of substance abuse treatment·2022

Area of Science:

  • Statistics
  • Biostatistics
  • Econometrics

Background:

  • Generalized estimating equations (GEE) are widely used for clustered data analysis.
  • The traditional sandwich standard error estimator often underestimates variability with small cluster counts.
  • This leads to inflated Type I error rates in hypothesis testing.

Purpose of the Study:

  • To develop improved methods for inference in GEE models.
  • To address the underestimation of variability by the sandwich estimator.
  • To enhance the accuracy of hypothesis tests for clustered data.

Main Methods:

  • Proposed bias-reduced linearization (BRL) to adjust the sandwich estimator.
  • Utilized Satterthwaite or saddlepoint approximations for reference distributions.

Related Experiment Videos

  • Conducted a large simulation study comparing various estimators and approximations.

    • Main Results:

    • The BRL method provided accurate variability estimates for fitted coefficients.
    • Tests using BRL achieved near-nominal Type I error rates when intra-cluster correlation (ICC) was small.
    • The proposed method outperformed traditional sandwich and other alternatives, particularly with small cluster numbers.

    Conclusions:

    • Bias-reduced linearization offers a superior approach to standard error estimation in GEE.
    • The new method improves hypothesis testing accuracy for clustered data.
    • Further research may explore performance with larger ICC values.