A minimum distance estimation approach to the two-sample location-scale problem.

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• The generalized Wilcoxon test is a standard method for comparing survival distributions.
• This test can fail when differences involve both location (shift) and scale (spread) parameters.
• Simultaneous location and scale changes obscure differences in lifetime distributions, as seen in mouse leukemia studies.

Purpose of the Study:

• To propose a novel statistical estimator for comparing lifetime distributions.
• To address the limitations of existing tests when location and scale parameters differ simultaneously.
• To provide a robust method for analyzing survival data, including right-censored observations.

Main Methods:

• Development of an estimator minimizing the average distance between quantile processes.
• Application of the estimator within a location-scale statistical model.
• Inclusion of methods for large-sample inference and handling right-censored data.

Main Results:

• The proposed estimator demonstrates improved ability to detect distributional differences under combined location-scale changes.
• The method is illustrated effectively using the classic mouse leukemia dataset.
• Theoretical large-sample properties of the estimator are established.

Conclusions:

• The new quantile process-based estimator offers a valuable alternative for survival data analysis.
• This approach enhances the detection of subtle differences in lifetime distributions.
• The method provides a more sensitive tool for biostatistical research, particularly in toxicology and longevity studies.