On symmetric semiparametric two-sample problem.

Area of Science:

• Statistics
• Biostatistics
• Statistical modeling

Background:

• Two-sample problems with symmetric distributions are common in biostatistics, particularly in case-control studies.
• Existing methods may not fully capture the nuances of symmetric data, necessitating specialized approaches.

Purpose of the Study:

• To propose a novel semiparametric model for analyzing two-sample data from symmetric distributions.
• To develop a robust estimation and hypothesis testing framework for symmetric data.

Main Methods:

• A semiparametric model is proposed, incorporating symmetry and a parametric form for the log ratio of densities.
• Maximum empirical likelihood estimation is employed for parameter estimation.
• A profile empirical likelihood ratio test is used for hypothesis testing.

Main Results:

• The proposed model effectively handles symmetric two-sample data, viewing it as a biased sampling problem with a symmetric constraint.
• The maximum empirical likelihood estimator exhibits degenerate Fisher information under the null hypothesis.
• The test statistic follows a mixture of chi-squared asymptotic distribution.

Conclusions:

• The developed semiparametric model and empirical likelihood methods provide a powerful tool for analyzing symmetric two-sample data.
• Simulation studies confirm the statistical power of the proposed methods, even with model misspecification.
• The methods are demonstrated on real-world examples, highlighting their practical applicability.