Permuting density-based empirical likelihood ratio test statistics

Abstract

In this article we study two permutation tests based on the density-based empirical likelihood ratio (DBELR) test statistics with application to paired and two-sample data. The DBELR tests have been extensively studied in statistical literature and are widely recognized for their distribution-free nature and high, stable power across various testing problems. The proposed methods enhance the use of DBELR tests in a more convenient manner that significantly reduces computation times and generates permutation p-values to make test decisions while maintaining the desired level α. An extensive Monte Carlo simulation study shows that the proposed methods inherit the stable and high-power property of the original DBELR tests. A real-world data example is used to demonstrate the applicability of the proposed methods.

Keywords:

• Density-based empirical likelihood ratio
• Level α test
• Paired and two-sample data
• Permutation test

Disclosure statement

No potential conflict of interest was reported by the author(s).

Acknowledgements

The authors thank two reviewers for their insightful and helpful comments that led to a substantial improvement in this article.

Additional information

Funding

The research is funded under the 2022-2023 Faculty Research, Scholarship, and Creativity Support Grant program from California State University, East Bay.