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 practical approach to computing power for generalized linear models with nominal, count, or ordinal responses.

Robert H Lyles1, Hung-Mo Lin, John M Williamson

  • 1Department of Biostatistics, The Rollins School of Public Health of Emory University, 1518 Clifton Rd. N.E., Atlanta, GA 30322, USA. rlyles@sph.emory.edu

Statistics in Medicine
|July 4, 2006
PubMed
Summary

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

Superficial Cervical Plexus Block and Quality of Recovery after Thyroidectomy: A Randomized Clinical Trial.

Anesthesiology·2026
Same author

Preoperative Olanzapine and Quality of Recovery after Ambulatory Surgery: A Randomized Clinical Trial.

Anesthesiology·2026
Same author

Glucocorticoid-Enhanced Fascial Plane and Peripheral Nerve Blocks Versus Periarticular and Local Infiltration Analgesia in Total Hip Arthroplasty: A Prospective Randomized Controlled Trial.

The Journal of bone and joint surgery. American volume·2026
Same author

Efficacy of a smartphone game to increase age and condom use at first sex among adolescents in Kenya (Tumaini): a randomised controlled trial.

The Lancet. Child & adolescent health·2026
Same author

Longitudinal Effects of a Smartphone Game (Tumaini) for HIV Prevention Among Kenyan Adolescents: 45-Month Trajectories of Condom Use-Related Proximal Outcomes From a Randomized Controlled Trial.

Journal of medical Internet research·2026
Same author

Changes in Use of Physical Restraints in Intensive Care Unit Patients During the COVID-19 Pandemic.

American journal of critical care : an official publication, American Association of Critical-Care Nurses·2026

Data analysts can now use a simple, unified method for power calculations in generalized linear models. This approach accurately estimates power for various outcomes and covariates, even with moderate sample sizes.

Area of Science:

  • Statistics
  • Biostatistics
  • Data Analysis

Background:

  • Power calculations are crucial for study design but existing methods for generalized linear models (GLMs) are often complex.
  • Current approaches for binary, ordinal, or count outcomes can be computationally demanding, limited with covariates, or lack validation for moderate sample sizes.

Purpose of the Study:

  • To present a straightforward and unified method for estimating conditional power in GLMs.
  • To offer a flexible approach that accommodates various outcome types and covariates, including continuous ones, without discretization.

Main Methods:

  • Fit a GLM to an expanded dataset using calculated weights representing response probabilities.
  • Utilize the variance-covariance matrix from the GLM fit with a non-central chi-square approximation for the Wald statistic.

Related Experiment Videos

  • Alternatively, re-fit the model under the null hypothesis to approximate power using the likelihood ratio statistic.

    • Main Results:

    • The proposed method is accurate for realistic sample sizes.
    • It demonstrates flexibility in handling one or more continuous covariates.
    • Simulations confirm the method's accuracy across various outcome types and covariate patterns.

    Conclusions:

    • This unified approach simplifies power calculations for GLMs, benefiting data analysts.
    • The method is computationally undemanding and highly flexible, particularly for complex covariate structures.
    • It provides a reliable tool for study design involving non-continuous outcomes.