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

Performance of weighted estimating equations for longitudinal binary data with drop-outs missing at random.

John S Preisser1, Kurt K Lohman, Paul J Rathouz

  • 1Department of Biostatistics, CB #7420, School of Public Health, University of North Carolina, Chapel Hill 27599, USA. jpreisse@bios.unc.edu

Statistics in Medicine
|October 9, 2002
PubMed
Summary

Weighted generalized estimating equations (GEE) reduce bias in longitudinal binary data analysis with missing at random drop-outs. However, misspecification of the drop-out model can lead to worse performance than standard GEE.

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

Analysis of Stepped-Wedge Cluster Randomized Trials: A Tutorial Using Marginal Models.

Statistics in medicine·2026
Same author

Corrigendum to "Establishing a training plan and estimating inter-rater reliability across the multi-site Texas Childhood Trauma Research Network" [Psychiatry Research 323 (2023) 115168].

Psychiatry research·2026
Same author

A Pharmacist Consultant Service for Deprescribing Opioids and Benzodiazepines in Older Adults: A Cluster Randomized Trial.

JAMA network open·2026
Same author

Impact of a web-based breast cancer surgery decision aid on knowledge and perceptions of feeling informed in clinics that care for socioeconomically disadvantaged patients: An Alliance Clinical Trial (A231701CD).

Cancer·2026
Same author

Corrigendum to "Clinical characteristics and longitudinal associations with obsessive-compulsive disorder in youth exposed to trauma" [J Mood Anxiety Disord 10C (2025) 100117].

Journal of mood and anxiety disorders·2026
Same author

Three open questions in polygenic score portability.

Nature communications·2026

Area of Science:

  • Biostatistics
  • Longitudinal Data Analysis
  • Statistical Modeling

Background:

  • Generalized Estimating Equations (GEE) are widely used for incomplete longitudinal binary data.
  • Missing at random (MAR) drop-outs can bias parameter estimates in standard GEE models for marginal means.
  • Weighted estimating equations offer a potential solution for consistent estimation under MAR.

Purpose of the Study:

  • To compare the performance of unweighted and weighted GEE methods.
  • To evaluate bias and efficiency in models for time-specific means of binary responses with MAR drop-outs.
  • To assess the impact of drop-out model misspecification on GEE performance.

Main Methods:

  • A simulation study was conducted to compare statistical approaches.

Related Experiment Videos

  • Unweighted and weighted GEE were applied to longitudinal binary data with MAR drop-outs.
  • Observation-level and cluster-level weights were investigated for the weighted GEE approach.

    • Main Results:

    • Weighted GEE demonstrated smaller finite sample bias compared to unweighted GEE.
    • Misspecification of the drop-out model led to instances where weighted GEE performed worse than standard GEE.
    • Observation-level weights in weighted GEE yielded more efficient estimates than cluster-level weights.

    Conclusions:

    • Weighted GEE is a valuable tool for reducing bias in longitudinal binary data analysis with MAR drop-outs when the drop-out model is correctly specified.
    • Careful specification of the drop-out mechanism is crucial for the reliable performance of weighted GEE.
    • Observation-level weighting provides superior efficiency in weighted GEE analyses.