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

Problems due to small samples and sparse data in conditional logistic regression analysis.

S Greenland1, J A Schwartzbaum, W D Finkle

  • 1Department of Epidemiology, School of Public Health, University of California at Los Angeles, USA.

American Journal of Epidemiology
|March 9, 2000
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

Glaucoma outcome studies using existing databases: opportunities and limitations.

Journal of glaucoma·2009
Same author

Small-sample bias and corrections for conditional maximum-likelihood odds-ratio estimators.

Biostatistics (Oxford, England)·2003
Same author

On the bias produced by quality scores in meta-analysis, and a hierarchical view of proposed solutions.

Biostatistics (Oxford, England)·2003
Same author

Ecologic versus individual-level sources of bias in ecologic estimates of contextual health effects.

International journal of epidemiology·2002
Same author

Sensitivity analysis, Monte Carlo risk analysis, and Bayesian uncertainty assessment.

Risk analysis : an official publication of the Society for Risk Analysis·2001
Same author

Putting background information about relative risks into conjugate prior distributions.

Biometrics·2001
Same journal

Correction to: Home dampness and molds and occurrence of respiratory tract infections in the first 27 years of life: the Espoo Cohort Study.

American journal of epidemiology·2026
Same journal

A SIMPLE AND POWERFUL TEST OF VACCINE WANING.

American journal of epidemiology·2026
Same journal

Association Between maternal body mass index, offspring growth and pubertal timing: results from a longitudinal birth cohort study.

American journal of epidemiology·2026
Same journal

Correction to: Developing a novel algorithm to identify incident and prevalent dementia in Medicare claims-the ARIC Study.

American journal of epidemiology·2026
Same journal

RE: advancing observational research on arts and health: theory-informed approaches using the RADIANCE framework.

American journal of epidemiology·2026
Same journal

Maternal Cesarean Section and Offspring ASD or ADHD Risk: A Nurses' Health Study II Analysis.

American journal of epidemiology·2026
See all related articles

Conditional logistic regression can be biased with sparse data, affecting epidemiological study results. Researchers suggest bias detection methods and alternative statistical approaches for more reliable inferences.

Area of Science:

  • Epidemiology
  • Biostatistics

Background:

  • Conditional logistic regression (CLR) addresses sparse-data biases in ordinary logistic regression.
  • However, CLR is a large-sample method susceptible to bias with infrequent matched sets or complex models.

Purpose of the Study:

  • To describe sparse-data bias issues in matched case-control studies.
  • To provide examples and discuss detection and mitigation strategies for these biases.

Main Methods:

  • Analysis of sparse-data bias in conditional logistic regression.
  • Case examples from studies on electrical wiring and childhood leukemia, and diet and glioma.
  • Exploration of bias detection through data inspection and sensitivity analyses.

Main Results:

Related Experiment Videos

  • Sparse-data bias can lead to erroneous conclusions regarding confounding, effect modification, and dose-response relationships.
  • Bias can interact with other sources of error in epidemiological analyses.
  • Problems are not limited to CLR but can affect any likelihood-based analysis.

    • Conclusions:

    • Sparse-data bias is a significant concern in matched case-control studies and other likelihood-based analyses.
    • Detection involves careful data review and sensitivity testing.
    • Alternative methods like Bayesian or empirical-Bayes approaches may offer greater robustness.