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

A covariance estimator for GEE with improved small-sample properties.

L A Mancl1, T A DeRouen

  • 1Department of Dental Public Health Sciences, University of Washington, Seattle 98195, USA. lman@biostat.washington.edu

Biometrics
|March 17, 2001
PubMed
Summary

A new bias-corrected covariance estimator improves hypothesis testing for generalized estimating equations (GEE) with small sample sizes and binary outcomes. This method maintains accurate test sizes, unlike traditional robust and jackknife estimators.

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

Our Essential and Endangered Dentist-Scientist Workforce.

JDR clinical and translational research·2019
Same author

Promises and Pitfalls in the Use of "Big Data" for Clinical Research.

Journal of dental research·2015
Same author

Implementation partnerships in a community-based intergenerational oral health study.

Community dental health·2015
Same author

Critique of reanalysis of Casa Pia data on associations of porphyrins and glutathione-S-transferases with dental amalgam exposure.

Human & experimental toxicology·2014
Same author

Emphasizing Relevance.

Journal of dental research·2014
Same author

Dental hygiene work in a clinical trial.

International journal of dental hygiene·2008

Area of Science:

  • Biostatistics
  • Statistical Modeling
  • Longitudinal Data Analysis

Background:

  • Generalized estimating equations (GEE) are widely used for analyzing longitudinal or clustered data.
  • Traditional robust covariance estimators in GEE can lead to inflated Type I error rates (inflated size) with small numbers of independent clusters.
  • Resampling methods like jackknife and bootstrap are alternatives but can fail with small sample sizes and binary responses due to zero or near-zero cell counts.

Purpose of the Study:

  • To propose a novel bias-corrected covariance estimator for GEE that overcomes the limitations of existing methods, particularly for small sample sizes and binary data.
  • To evaluate the performance of the proposed estimator against standard robust and jackknife estimators in hypothesis testing scenarios.

Main Methods:

Related Experiment Videos

  • Development of a bias-corrected covariance estimator designed to avoid breakdown issues associated with resampling methods.
  • A simulation study comparing the proposed estimator with robust and jackknife estimators.
  • The simulation varied the number of subjects (10-40) and cluster sizes (16-64), including equal and unequal cluster sizes, for binary responses.

    • Main Results:

    • The bias-corrected covariance estimator produced hypothesis tests with sizes close to the nominal level, even with as few as 10 subjects and unequal cluster sizes.
    • In contrast, the robust and jackknife covariance estimators resulted in inflated test sizes, sometimes 2-3 times the nominal level, under similar conditions.
    • The proposed method demonstrated superior performance in maintaining accurate statistical inference in small cluster settings.

    Conclusions:

    • The proposed bias-corrected covariance estimator offers a reliable alternative for hypothesis testing in GEE when dealing with a small number of clusters and binary outcomes.
    • This method effectively addresses the size inflation issues seen with traditional estimators in small sample scenarios.
    • The practical utility is demonstrated through an application to a randomized clinical trial for periodontal disease treatment.