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

Advanced statistics: statistical methods for analyzing cluster and cluster-randomized data.

Robert L Wears1

  • 1Department of Emergency Medicine, University of Florida Health Center, Jacksonville, FL 32209, USA. wears@ufl.edu

Academic Emergency Medicine : Official Journal of the Society for Academic Emergency Medicine
|April 3, 2002
PubMed
Summary

Cluster randomization in clinical research involves group-level allocation, not individual. Failing to account for this clustered data violates independence, leading to inaccurate statistical results and sample size calculations.

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

Communication in the Electronic Age: an Analysis of Face-to-Face Physician-Nurse Communication in the Emergency Department.

Journal of healthcare informatics research·2022
Same author

Triage, Machine Learning, Algorithms, and Becoming the Borg.

Annals of emergency medicine·2018
Same author

The Secret Life of Policies.

Annals of emergency medicine·2017
Same author

Clinical Decision Aids or Clinician Gestalt? Hard to Know Which Is Better: September 2017 Annals of Emergency Medicine Journal Club.

Annals of emergency medicine·2017
Same author

Understanding Emergency Care Delivery Through Computer Simulation Modeling.

Academic emergency medicine : official journal of the Society for Academic Emergency Medicine·2017
Same author

The Rush from Judgment.

Annals of emergency medicine·2017

Area of Science:

  • Biostatistics
  • Health Services Research
  • Clinical Trials

Background:

  • Cluster randomization allocates interventions to groups, not individuals, common in health services research.
  • Naturally occurring groups (e.g., neighborhoods) in observational studies also exhibit within-cluster similarity.
  • Observations within clusters are more alike than random selections, violating statistical independence assumptions.

Purpose of the Study:

  • Introduce the challenges of analyzing clustered data in clinical research.
  • Provide guidance on methods for analyzing clustered data.
  • Offer strategies for calculating appropriate sample sizes for studies with clustered data.

Main Methods:

  • Discusses the implications of ignoring within-cluster dependence in statistical analysis.

Related Experiment Videos

  • Highlights the impact on p-values, confidence intervals, and sample size estimations.
  • Draws parallels to the 'unit of analysis error' from repeated measures.

    • Main Results:

    • Failure to account for clustered data leads to underestimated p-values and overly narrow confidence intervals.
    • Sample size estimations are significantly reduced when clustering is not considered.
    • This can dramatically impact study design and the validity of research findings.

    Conclusions:

    • Emphasizes the critical need to address clustered data in clinical research design and analysis.
    • Recommends specific statistical methods and sample size calculations for clustered data.
    • Advises on statistical software and general principles for planning, analyzing, and presenting studies with clustered data.