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

Comparison of Bayesian model averaging and stepwise methods for model selection in logistic regression.

Duolao Wang1, Wenyang Zhang, Ameet Bakhai

  • 1Department of Epidemiology and Population Health, Medical Statistics Unit, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK. duolao.wang@lshtm.ac.uk

Statistics in Medicine
|October 27, 2004
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

Drug-Coated Balloons Versus Drug-Eluting Stents for Patients With Long De Novo Coronary Artery Lesions: Insights From the REC-CAGEFREE I Trial.

Catheterization and cardiovascular interventions : official journal of the Society for Cardiac Angiography & Interventions·2026
Same author

Pro/synbiotics reduce systemic inflammation and improve gut health in Kenyan infants: An open label, randomized, 4-arm, phase II trial.

Cell reports. Medicine·2026
Same author

Effect of renin-angiotensin system inhibition on left ventricular mass regression after transcatheter aortic valve replacement: a randomised controlled trial.

Heart (British Cardiac Society)·2026
Same author

Effect of Left Ventricular Ejection Fraction on Outcomes With Drug-coated Balloons vs Drug-Eluting Stents for De Novo Coronary Artery Disease.

The Canadian journal of cardiology·2026
Same author

Neutrophil Extracellular Traps Are Associated With Early Neurological Deterioration in Patients With Acute Ischemic Stroke Receiving Intravenous Thrombolysis: A Prospective Cohort Study.

Journal of the American Heart Association·2026
Same author

Intravenous Tirofiban After Tenecteplase in Acute Ischemic Stroke: The INSTANT Randomized Clinical Trial.

JAMA·2026

Bayesian model averaging is a superior method for logistic regression, outperforming stepwise variable selection in disease prediction. This approach accounts for model uncertainty, leading to more accurate inferences and predictions.

Area of Science:

  • Biostatistics
  • Epidemiology
  • Medical Informatics

Background:

  • Logistic regression is standard for disease predictor analysis.
  • Stepwise variable selection in logistic regression ignores unselected variables and selection uncertainty.
  • This limitation can impact the accuracy of predictor inference and disease risk assessment.

Purpose of the Study:

  • To compare Bayesian model averaging (BMA) with stepwise logistic regression for variable selection.
  • To evaluate the performance of BMA versus stepwise methods in prediction accuracy.
  • To assess the utility of BMA in handling predictor uncertainty in logistic regression models.

Main Methods:

  • Simulated datasets were used to compare BMA and stepwise logistic regression.
  • The Framingham Heart Study dataset was utilized for real-world validation.

Related Experiment Videos

  • Inferences and predictions were made using posterior model probabilities in BMA.

    • Main Results:

    • Bayesian model averaging (BMA) identified the correct predictor model in most simulations.
    • BMA demonstrated superior predictive performance compared to stepwise logistic regression.
    • The study highlights BMA's effectiveness in addressing variable selection uncertainty.

    Conclusions:

    • Bayesian model averaging (BMA) offers a more robust approach to logistic regression than stepwise methods.
    • BMA improves the accuracy of disease prediction by accounting for model uncertainty.
    • The findings support the adoption of BMA for reliable predictor analysis in epidemiological studies.