Inference on the overlap coefficient: The binormal approach and alternatives.

Area of Science:

• Statistics
• Biostatistics

Background:

• The overlap coefficient quantifies similarity between two distributions.
• Accurate confidence interval methods are needed for applied researchers.
• Existing methods lack software and broad applicability.

Purpose of the Study:

• Develop accurate parametric and non-parametric methods for overlap coefficient confidence intervals.
• Provide R-code for implementation.
• Address limitations in current literature.

Main Methods:

• Developed parametric and non-parametric inferential procedures.
• Utilized the binormal model for parametric approaches.
• Assessed methods via a large simulation study.

Main Results:

• Parametric methods, especially the binormal model, demonstrated superior performance.
• The developed methods are suitable for diverse distributional scenarios.
• R-code facilitates practical implementation.

Conclusions:

• The study provides robust statistical methods for the overlap coefficient.
• Parametric approaches offer reliable confidence intervals for distribution similarity.
• Findings are applicable to fields like cognitive function assessment.