A Permutation Approach to Testing Interactions for Binary Response by Comparing Correlations Between Classes

DOI10.1080/01621459.2014.993079zbMATH Open1373.62278arXiv1206.6519OpenAlexW2018494427MaRDI QIDQ5367480FDOQ5367480

Authors: Noah Simon, Robert Tibshirani 

Publication date: 13 October 2017

Published in: Journal of the American Statistical Association (Search for Journal in Brave)

Abstract: To date, testing interactions in high dimensions has been a challenging task. Existing methods often have issues with sensitivity to modeling assumptions and heavily asymptotic nominal p-values. To help alleviate these issues, we propose a permutation-based method for testing marginal interactions with a binary response. Our method searches for pairwise correlations which differ between classes. In this manuscript, we compare our method on real and simulated data to the standard approach of running many pairwise logistic models. On simulated data our method finds more significant interactions at a lower false discovery rate (especially in the presence of main effects). On real genomic data, although there is no gold standard, our method finds apparent signal and tells a believable story, while logistic regression does not. We also give asymptotic consistency results under not too restrictive assumptions.

Full work available at URL: https://arxiv.org/abs/1206.6519

Recommendations

• Comparison of permutation methods for the partial correlation and partial mantel tests
• Testing equality of correlation coefficients in two populations via permutation methods
• On the permutation test in canonical correlation analysis
• Learning Theory
• Interaction testing: residuals-based permutations and parametric bootstrap in continuous, count, and binary data
• Permuiation tesis for k-sample binomial data with comparisons of exact and approximate P-levels
• The classification permutation test: a flexible approach to testing for covariate imbalance in observational studies

zbMATH Keywords

correlationfalse discovery ratehigh dimensionallogistic regressionpermutation approach

Mathematics Subject Classification ID

Hypothesis testing in multivariate analysis (62H15) Measures of association (correlation, canonical correlation, etc.) (62H20) Generalized linear models (logistic models) (62J12)

Cited In (8)

• Penalized interaction estimation for ultrahigh dimensional quadratic regression
• Robust Variable and Interaction Selection for Logistic Regression and General Index Models
• Convex hierarchical testing of interactions
• Interaction testing: residuals-based permutations and parametric bootstrap in continuous, count, and binary data
• Test for conditional independence with application to conditional screening
• Letter to the editor: ``Exact permutation tests for correlated binomial variables by J. Yu, J. L. Kepner and R. Iyer
• Permutation Multiple Tests of Binary Features Do Not Uniformly Control Error Rates
• Ensemble survival tree models to reveal pairwise interactions of variables with time-to-events outcomes in low-dimensional setting

This page was built for publication: A Permutation Approach to Testing Interactions for Binary Response by Comparing Correlations Between Classes

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5367480)