Related Experiment Videos
Measuring effect size: a non-parametric analogue of omega 2
1Department of Psychology, University of Southern California, Los Angeles 90089-1061, USA.
Summary
This study introduces a non-parametric effect size measure as an alternative to omega 2. The .632 bootstrap estimator demonstrated superior performance in simulations for estimating this new measure.
Area of Science:
- Statistics
- Psychometrics
- Quantitative Psychology
Background:
- Omega 2 is a common effect size measure for comparing two groups.
- Omega 2 is sensitive to variance and can be misleading.
- A non-parametric alternative is needed for robust effect size estimation.
Purpose of the Study:
- To propose a non-parametric analogue of omega 2.
- To evaluate estimators for this new effect size measure.
- To address the limitations of traditional omega 2.
Main Methods:
- Developed a novel non-parametric effect size measure.
- Proposed four estimators for the new measure.
- Conducted a simulation study to compare estimator performance.
Main Results:
- The .632 bootstrap estimator showed the best performance.
- This estimator exhibited the lowest bias and mean squared error.
- Simulation results contrasted with initial expectations.
Conclusions:
- The proposed non-parametric effect size measure offers a robust alternative.
- The .632 bootstrap estimator is recommended for practical application.
- This work advances effect size estimation in statistical comparisons.