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

Confidence intervals for multinomial logistic regression in sparse data.

Shelley B Bull1, Juan Pablo Lewinger, Sophia S F Lee

  • 1Samuel Lunenfeld Research Institute, Prosserman Centre for Health Research, Mount Sinai Hospital, Toronto, Ont., Canada M5G 1X5. bull@mshri.on.ca

Statistics in Medicine
|February 21, 2006
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

TGF-β Pathway-Based Polygenic Risk Score Modifies the Association between Red Meat Intake and Colorectal Cancer Risk: Application of a Novel Pathway-Based PRS Method.

Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology·2026
Same author

Accommodating the Analysis Model in Multiple Imputation for the Weibull Mixture Cure Model: Performance Under Penalized Likelihood.

Statistics in medicine·2026
Same author

Genetic risk factors modulate the association between physical activity and colorectal cancer.

BMC medicine·2026
Same author

Plasma Metabolomic and Proteomic Signatures of Blood Pressure Management After Bariatric Surgery Among Adolescents.

Hypertension (Dallas, Tex. : 1979)·2025
Same author

Using gene-environment interactions to explore pathways for colorectal cancer risk.

EBioMedicine·2025
Same author

Genetic risk factors modulate the association between physical activity and colorectal cancer.

Research square·2025
Same journal

A Mixture of Distributed Lag Non-Linear Models to Account for Spatially Heterogeneous Exposure-Lag-Response Associations.

Statistics in medicine·2026
Same journal

Practical Considerations for Gaussian Process Modeling for Causal Inference in Quasi-Experimental Studies With Panel Data.

Statistics in medicine·2026
Same journal

Covariate Adjustment for Wilcoxon Two Sample Statistic and Test.

Statistics in medicine·2026
Same journal

Beyond Fixed Thresholds: Optimizing Summaries of Wearable Device Data via Piecewise Linearization of Quantile Functions.

Statistics in medicine·2026
Same journal

A Causal Framework for Evaluating the Total Effect of Strategies Aiming to Expand Screening and to Improve Outcomes.

Statistics in medicine·2026
Same journal

Causal Effects on Nonterminal Event Time With Application to Antibiotic Usage and Future Resistance.

Statistics in medicine·2026
See all related articles

Penalized maximum likelihood estimates (PLEs) offer a robust alternative to conventional maximum likelihood estimates (MLEs) for logistic regression in small or sparse samples. Profile confidence intervals for PLEs are recommended over MLE methods, especially when dealing with data separation.

Area of Science:

  • Statistics
  • Biostatistics
  • Statistical Modeling

Background:

  • Conventional maximum likelihood estimation (MLE) in logistic regression can be unreliable with small or sparse sample sizes.
  • Data separation and finite sample bias can lead to infinite or unstable MLEs, necessitating alternative approaches.
  • Penalized maximum likelihood estimation (PLE) offers a viable solution by modifying the score function to mitigate bias.

Purpose of the Study:

  • To develop and evaluate methods for constructing confidence intervals (CIs) in penalized multinomial logistic regression models.
  • To compare the performance of PLE-based CIs against conventional MLE-based CIs in terms of coverage and length.
  • To provide recommendations for CI selection in logistic regression, particularly for challenging datasets.

Main Methods:

Related Experiment Videos

  • Modification of the logistic regression score function using Jeffreys prior to obtain penalized maximum likelihood estimates (PLEs).
  • Development of confidence interval construction methods for PLEs in multinomial logistic regression.
  • Simulation studies comparing profile CIs and asymptotic Wald-type intervals for PLEs against MLE-based methods in trinomial logistic regressions with binary and continuous covariates.

Main Results:

  • Penalized maximum likelihood estimates (PLEs) always exist, even when maximum likelihood estimates (MLEs) are infinite, making them suitable for sparse data.
  • Profile confidence intervals (CIs) for PLEs demonstrated superior coverage and length compared to asymptotic Wald-type intervals in simulation studies.
  • PLE profile CIs were preferred over MLE methods when finite sample bias and data separation were present.

Conclusions:

  • Penalized maximum likelihood estimation provides a reliable alternative to conventional maximum likelihood estimation for logistic regression, especially in small or sparse datasets.
  • Profile confidence intervals are recommended for penalized maximum likelihood estimates due to their robust performance.
  • The proposed PLE profile CI methods offer a preferred approach over traditional MLE methods when dealing with data separation and finite sample bias in logistic regression analysis.