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

Maximally selected chi2 statistics for k x 2 tables.

R A Betensky1, D Rabinowitz

  • 1Department of Statistics, Columbia University, New York, New York 10032, USA. betensky@sdac.harvard.edu

Biometrics
|April 25, 2001
PubMed
Summary

Maximally selecting thresholds for dichotomized outcomes in epidemiological studies can lead to incorrect statistical tests. This research derives accurate asymptotic distributions for these maximally selected chi-squared statistics, improving analysis reliability.

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

Accounting for incomplete testing in the estimation of epidemic parameters.

medRxiv : the preprint server for health sciences·2020
Same author

Nonidentifiability in the presence of factorization for truncated data.

Biometrika·2019
Same author

Hypothesis Tests for Neyman's Bias in Case-Control Studies.

Journal of applied statistics·2018
Same author

Preoperative angiography and external carotid artery embolization of juvenile nasopharyngeal angiofibromas in a tertiary referral paediatric centre.

Clinical radiology·2013
Same author

Adhesion molecules, endothelin-1 and lung function in seven population-based cohorts.

Biomarkers : biochemical indicators of exposure, response, and susceptibility to chemicals·2013
Same author

Gain of chromosome arm 1q in atypical meningioma correlates with shorter progression-free survival.

Neuropathology and applied neurobiology·2011

Area of Science:

  • Epidemiology
  • Biostatistics
  • Clinical Trials

Background:

  • Continuous outcomes are frequently dichotomized using thresholds in epidemiological analyses.
  • Standard chi-squared (χ²) statistics are used to test for associations or trends in proportions across strata.
  • Selecting thresholds to maximize test statistics invalidates the nominal chi-squared reference distributions.

Purpose of the Study:

  • To derive the asymptotic distributions for maximally selected chi-squared statistics for association and trend.
  • To address the issue of incorrect statistical inference when thresholds are optimized for significance.
  • To provide a methodologically sound approach for analyzing dichotomized outcomes in epidemiological data.

Main Methods:

  • Derivation of asymptotic distributions for maximally selected chi-squared statistics.

Related Experiment Videos

  • Application of the methodology to a k x 2 contingency table.
  • Illustration using data from a clinical trial for Acquired Immunodeficiency Syndrome (AIDS).

    • Main Results:

    • The study successfully derived the asymptotic distributions for the maximally selected chi-squared statistics.
    • Simulation experiments were conducted to evaluate the accuracy of these distributions in moderate sample sizes.
    • The derived distributions provide a correct reference for hypothesis testing in such scenarios.

    Conclusions:

    • The derived asymptotic distributions enable accurate statistical testing when thresholds are chosen to maximize chi-squared statistics.
    • This methodology enhances the reliability of epidemiological analyses involving dichotomized outcomes.
    • The findings are validated through simulations and an AIDS clinical trial example.