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

Tests against qualitative interaction: exact critical values and robust tests.

M J Silvapulle1

  • 1Department of Statistical Science, La Trobe University, Bundoora, Australia. m.silvapulle@latrobe.edu.au

Biometrics
|January 5, 2002
PubMed
Summary

This study provides exact finite sample critical values for testing qualitative interaction between treatments and patient groups. Robust tests are also developed for improved power when outliers are present.

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

Ladders: accidents waiting to happen.

The Medical journal of Australia·2001
Same author

Epileptology of the first-seizure presentation: a clinical, electroencephalographic, and magnetic resonance imaging study of 300 consecutive patients.

Lancet (London, England)·1998
Same author

On tests against one-sided hypotheses in some generalized linear models.

Biometrics·1994
Same author

Epidermal growth factor alters the electrolyte profile of lactating ewes (Ovis aries).

Comparative biochemistry and physiology. Comparative physiology·1992
Same author

Fluid balance, electrolyte profiles and plasma parathyroid hormone concentrations in ewes treated with epidermal growth factor.

The Journal of endocrinology·1992
Same author

On testing for threshold values.

Biometrics·1991

Area of Science:

  • Biostatistics
  • Clinical Trial Design

Background:

  • Qualitative interaction occurs when a treatment benefits one patient group while harming another.
  • Gail and Simon (1985) proposed a large-sample test for this interaction, widely recognized in statistical literature.
  • Existing methods lack exact finite sample results, particularly for non-ideal error distributions.

Purpose of the Study:

  • To derive exact finite sample critical values for testing qualitative interaction under a normal error distribution.
  • To develop power-robust tests for qualitative interaction, especially when long-tailed error distributions or outliers are anticipated.
  • To compare the performance of exact versus asymptotic critical values and the benefits of robust tests.

Main Methods:

  • Derivation of exact finite sample results for a test statistic based on the F-ratio.

Related Experiment Videos

  • Calculation of critical values for various sample sizes and significance levels.
  • Development of robust tests using M-estimators instead of least squares estimators.
  • Simulation studies to compare exact critical values with asymptotic ones and to evaluate the power of robust tests.

    • Main Results:

    • A table of exact critical values for the normal error distribution is provided.
    • Exact critical values are shown to be preferable to asymptotic values across various error distributions.
    • Robust tests using M-estimators demonstrate substantial power advantages over standard tests when outliers are present.
    • The efficiency robustness of M-estimators translates directly to power robustness in hypothesis testing.

    Conclusions:

    • The derived exact finite sample critical values offer improved accuracy for testing qualitative interaction with normal errors.
    • The proposed robust tests are recommended when the presence of outliers is suspected, offering superior power.
    • These findings enhance the statistical toolkit for analyzing treatment effects across different patient subgroups.