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

Sample size determinations using logistic regression with pilot data

V F Flack1, T L Eudey

  • 1UCLA Department of Biostatistics, 90024-1772.

Statistics in Medicine
|June 15, 1993
PubMed
Summary

This study proposes a logistic regression model for sample size calculation in prediction studies. This method offers more realistic and smaller prediction errors compared to traditional binomial approaches, potentially making studies feasible.

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

Findings of clinical and radiographic caries among several adult age groups.

Oral surgery, oral medicine, oral pathology, oral radiology, and endodontics·1998
Same author

The reproducibility of muscle and joint tenderness detection methods and maximum mandibular movement measurement for the temporomandibular system.

Journal of orofacial pain·1998
Same author

Waking and sleeping temporalis EMG levels in tension-type headache patients.

Journal of orofacial pain·1998
Same author

Relationships between clinician variability and radiographic guidelines.

Journal of dental research·1996
Same author

The analysis of incomplete data in the three-period two-treatment cross-over design for clinical trials.

Statistics in medicine·1996
Same author

Cross-correlation between stress, pain, physical activity, and temporalis muscle EMG in tension-type headache.

Cephalalgia : an international journal of headache·1995

Area of Science:

  • Biostatistics
  • Epidemiology
  • Health Services Research

Background:

  • Estimating prediction proportions (p) is crucial in studies where covariates influence outcomes.
  • Pilot data suggest covariates significantly impact prediction proportions and can be modeled using logistic regression.
  • Traditional binomial sample size calculations may not be optimal when covariates are influential.

Purpose of the Study:

  • To develop and validate a logistic regression-based method for sample size justification in projected studies.
  • To compare the efficiency and feasibility of logistic-based versus binomial-based sample size calculations.
  • To provide a more scientifically realistic approach to sample size determination when covariates are important.

Main Methods:

  • Utilizing pilot data to establish a logistic regression model for the prediction proportion (p).

Related Experiment Videos

  • Deriving sample size requirements based on the parameters of the fitted logistic model.
  • Comparing prediction standard errors and sample size requirements between the proposed logistic method and the standard binomial method.
  • Illustrating the methodology with a case study using dental radiograph pilot data.

    • Main Results:

    • Logistic regression-based sample sizes are more scientifically reasonable and yield smaller prediction standard errors than binomial methods for the same sample size.
    • The proposed logistic approach can enhance the feasibility of research proposals by optimizing sample size.
    • The dental radiograph example demonstrates the practical application and benefits of the logistic-based sample size calculation.

    Conclusions:

    • A logistic regression model provides a more realistic and efficient basis for sample size calculation than the traditional binomial approach when covariates are influential.
    • This method can lead to more feasible and scientifically robust study designs.
    • Researchers should consider logistic-based sample size calculations for studies involving prediction proportions and influential covariates.