TABLE 3.
Envelope test and rank test (RT) with partial differences (pro = 10%)
| Simulation scenario | Outlier prob. | T 1 | T 2 | T 3 | T 4 | T 5 | T 6 | RT |
|---|---|---|---|---|---|---|---|---|
| Scenario 1 | θ = 0 | 0.084 | 0.082 | 0.062 | 0.070 | 0.070 | 0.082 | 0.054 |
| θ = 0.05 | 0.066 | 0.102 | 0.088 | 0.098 | 0.142 | 0.700 | 0.084 | |
| Scenario 2 | θ = 0 | 0.174 | 0.572 | 0.718 | 0.756 | 0.876 | 0.966 | 0.044 |
| θ = 0.05 | 0.122 | 0.438 | 0.544 | 0.642 | 0.678 | 0.674 | 0.084 | |
| Scenario 3 | θ = 0 | 0.486 | 0.980 | 0.998 | 1.000 | 1.000 | 1.000 | 0.124 |
| θ = 0.05 | 0.362 | 0.912 | 0.994 | 0.998 | 1.000 | 0.712 | 0.076 | |
| Scenario 4 | θ = 0 | 0.67 0 | 0.998 | 1.000 | 1.090 | 1.000 | 1.000 | 0.234 |
| θ = 0.05 | 0.544 | 0.988 | 1.000 | 1.000 | 1.000 | 0.776 | 0.076 | |
| Scenario 5 | θ = 0 | 0.750 | 0.996 | 1.000 | 1.000 | 1.000 | 1.000 | 0.338 |
| θ = 0.05 | 0.628 | 0.976 | 1.000 | 1.000 | 1.000 | 0.918 | 0.110 |
Notes: Results for five simulation scenarios analogous to the ones described above where the differences between control and test groups are only in pro = 10% of the domain. Numbers represent proportion of times (out of 500 iterations) a test rejected the null hypothesis that control and test subjects have the same population median. Results are based on the envelope tests based on test statistics T1, T2, T3, T4, T5, T6 and on the rank test RT.