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 permutation test for inference in logistic regression with small- and moderate-sized data sets.

Douglas M Potter1

  • 1Biostatistics Department, Graduate School of Public Health, and Biostatistics Facility, University of Pittsburgh Cancer Institute, University of Pittsburgh, Suite 325, Sterling Plaza, 201 North Craig Street, Pittsburgh, PA 15213, USA. potter@upci.pitt.edu

Statistics in Medicine
|October 30, 2004
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

Induction of robust type-I CD8+ T-cell responses in WHO grade 2 low-grade glioma patients receiving peptide-based vaccines in combination with poly-ICLC.

Clinical cancer research : an official journal of the American Association for Cancer Research·2014
Same author

Antigen-specific immune responses and clinical outcome after vaccination with glioma-associated antigen peptides and polyinosinic-polycytidylic acid stabilized by lysine and carboxymethylcellulose in children with newly diagnosed malignant brainstem and nonbrainstem gliomas.

Journal of clinical oncology : official journal of the American Society of Clinical Oncology·2014
Same author

Phase II study of induction fixed-dose rate gemcitabine and bevacizumab followed by 30 Gy radiotherapy as preoperative treatment for potentially resectable pancreatic adenocarcinoma.

Annals of surgical oncology·2013
Same author

LOH in the HLA class I region at 6p21 is associated with shorter survival in newly diagnosed adult glioblastoma.

Clinical cancer research : an official journal of the American Association for Cancer Research·2013
Same author

MUC1 vaccine for individuals with advanced adenoma of the colon: a cancer immunoprevention feasibility study.

Cancer prevention research (Philadelphia, Pa.)·2012
Same author

Pitfalls in retrospective analyses of biomarkers: a case study with metastatic melanoma patients.

Journal of immunological methods·2012

For small data sets, a simple permutation test offers accurate inference in logistic regression, unlike asymptotic methods. This permutation test controls type I error effectively, providing reliable results where other methods fail.

Area of Science:

  • Statistics
  • Biostatistics
  • Computational Statistics

Background:

  • Large sample inference for logistic regression is unreliable with small data sets.
  • Maximum likelihood estimates may not exist, invalidating standard asymptotic results.
  • Exact conditional logistic regression can be overly conservative and sensitive to data grouping.

Purpose of the Study:

  • To introduce a permutation test for logistic regression in small data settings.
  • To evaluate the performance of the permutation test against asymptotic and exact methods.
  • To provide a robust inferential method when maximum likelihood estimates fail.

Main Methods:

  • A permutation test was developed by replacing the variable of interest with residuals.
  • Logistic regressions were performed on permuted residuals to compute p-values.

Related Experiment Videos

  • Simulations involved binary outcome data with binary or lognormal independent variables.

    • Main Results:

    • The permutation test effectively controlled type I error in small (n=20) data sets, unlike the asymptotic likelihood ratio test.
    • In large (n=1000) data sets, the permutation test performed comparably to the asymptotic test.
    • Permutation test p-values were similar to mid-p-values from exact conditional logistic regression in small data.

    Conclusions:

    • A simple permutation test is a reliable alternative for logistic regression with small data.
    • This method overcomes limitations of existing approaches, including non-existent maximum likelihood estimates.
    • The permutation test provides accurate and robust inference across various data scenarios.