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 simulation study of sample size for multilevel logistic regression models.

Rahim Moineddin1, Flora I Matheson, Richard H Glazier

  • 1Department of Public Health Sciences, University of Toronto, Toronto, Canada. rahim.moineddin@utoronto.ca

BMC Medical Research Methodology
|July 20, 2007
PubMed
Summary
This summary is machine-generated.

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

Providing Mental Health and Substance Use Services to People Released From Custody During the COVID-19 Pandemic: A Qualitative Study of Service Providers.

The International journal of health planning and management·2026
Same author

Pandemic perspectives on mental health and substance use services for people released from custody: a focus on service providers.

International journal of prison health·2026
Same author

Clarifying the evidence and implications of prioritizing extended-release buprenorphine for opioid agonist treatment in Canada's federal correctional system.

Canadian journal of public health = Revue canadienne de sante publique·2026
Same author

Mitigating child health inequalities through equity, diversity, inclusion, and accessibility school practices in Canada.

Health promotion international·2025
Same author

Overall and substance use-specific healthcare utilization among individuals with and without criminal justice involvement in Ontario, Canada.

International journal of prison health·2025
Same author

How Problem Gambling Shapes Activities of Daily Living: A Narrative Inquiry.

The American journal of occupational therapy : official publication of the American Occupational Therapy Association·2024

For multilevel logistic regression models, aim for at least 50 groups with 50 individuals each for accurate estimates. Low prevalence events require larger sample sizes, ensuring more than one event per group.

Area of Science:

  • Health and social sciences research
  • Statistical modeling

Background:

  • Many health and social science studies feature hierarchical data structures (e.g., patients within physicians).
  • Binary outcomes (e.g., disease presence/absence) are common in these multilevel studies.
  • Determining adequate sample size for accurate multilevel model estimates is crucial.

Purpose of the Study:

  • To assess how individual and group sample sizes impact the accuracy of multilevel logistic regression models.
  • To examine the influence of outcome prevalence and intra-class correlation coefficient (ICC) on model estimates.

Main Methods:

  • Simulation studies were conducted to evaluate parameter and variance component estimation.
  • Sample sizes at both individual and group levels were systematically varied.

Related Experiment Videos

  • The effect of outcome prevalence and ICC was analyzed.

    • Main Results:

    • Fixed effect parameter estimates are unbiased with ≥100 groups and ≥50 individuals per group.
    • Variance component estimates show slight bias even with 100 groups and 50 individuals per group.
    • Biases are severe with a group size of 5; standard errors for variance components are underestimated.

    Conclusions:

    • A minimum of 50 groups with 50 individuals per group is recommended for valid multilevel logistic regression estimates.
    • For low prevalence events, increase group size to ensure >1 expected event per group.
    • Adjust sample size based on outcome prevalence for reliable results.