Related Experiment Videos
Sample size determination based on rank tests in clinical trials.
Hansheng Wang1, Bin Chen, Shein-Chung Chow
1Guanghua School of Management, Peking University, Beijing, PR China.
Journal of Biopharmaceutical Statistics
|October 31, 2003
Summary
This study provides new formulas for sample size determination using non-parametric rank tests like Wilcoxon's. The derived methods are effective for moderate sample sizes, improving statistical power calculations.
Area of Science:
- Statistics
- Non-parametric statistics
Background:
- Sample size determination is crucial for statistical power.
- Non-parametric tests are widely used when data assumptions are violated.
Purpose of the Study:
- To develop explicit formulas for sample size determination.
- To enhance the power function analysis for rank-based tests.
- To provide accurate sample size calculations for Wilcoxon's rank sum tests and independence tests.
Main Methods:
- Derivation of explicit formulas for test statistic variability under alternative hypotheses.
- Obtaining closed-form power functions using higher-order polynomial equations.
- Conducting simulation studies to evaluate finite sample performance.
Main Results:
- Novel formulas for sample size determination were derived.
- Closed-form power functions were obtained for rank-based tests.
- Simulation results confirmed the effectiveness of the derived methods for moderate sample sizes.
Conclusions:
- The derived sample size formulas are reliable for practical applications.
- The study enhances the theoretical foundation of non-parametric test power analysis.
- Accurate sample size determination is achievable with the proposed methods.