Skip to main content

Advertisement

Table 2 Relative power of t-test and Mann-Whitney given as relative efficiency. Values less than 1 indicate greater power of t-test; greater than 1 indicates superiority of Mann-Whitney. Results are combined across sample sizes and correlations.

From: Parametric versus non-parametric statistics in the analysis of randomized trials with non-normally distributed data

Distribution Post-treatment scores Change scores
Moderate positive skew: shift 0.9348 0.9835
Moderate positive skew: ratio 1.1382 1.0436
Moderate negative skew: shift 1.1833 1.0187
Moderate negative skew: ratio 0.9301 0.9825
Uniform: shift 0.9339 0.9846
Uniform: ratio 0.9488 0.9929
Extreme negative skew: shift 1.3769 1.1140
Extreme negative skew: ratio 1.6675 1.2046
Extreme asymmetry: shift 7.1461 0.5370
Extreme asymmetry: ratio 9.0091 0.6432
Normal: shift 0.9660 0.9726
Normal: ratio 0.9740 0.9760