Measuring Distribution Similarities Between Samples: A Distribution-Free Overlapping Index.

Area of Science:

• Cognitive and experimental psychology
• Statistical analysis
• Psychological research methodology

Background:

• Traditional statistical measures (t-statistic, Cohen's d) quantify sample differences but often rely on strict distributional assumptions (e.g., symmetry, unimodality).
• These assumptions can limit the validity and interpretability of results in psychological research.
• There is a need for robust statistical methods that are less sensitive to distributional assumptions.

Purpose of the Study:

• To introduce and illustrate a distribution-free overlapping measure as an alternative for quantifying sample differences.
• To demonstrate how this measure can be used to assess research hypotheses in terms of Bayesian evidence.
• To enhance the interpretability and reliability of data analysis in psychological research.

Main Methods:

• Utilized a distribution-free overlapping index to quantify differences between samples.
• Applied the index in three empirical applications within psychological research.
• Assessed research hypotheses using Bayesian evidence in conjunction with the overlapping measure.

Main Results:

• The overlapping index provides a viable, distribution-free alternative to traditional measures for quantifying sample differences.
• Empirical applications demonstrated the measure's utility in assessing hypotheses and improving data interpretation.
• The proposed method enhances the reliability of conclusions drawn from psychological studies.

Conclusions:

• The distribution-free overlapping measure offers a valuable tool for psychological researchers, overcoming limitations of traditional statistics.
• This approach improves the interpretability of data analysis and the confidence in research conclusions.
• Adoption of this measure can lead to more robust and reliable findings in experimental and applied psychology.