Table 2.
Comparison of the required sample size (power) per group (n1 = n2) for the three tests with 90% nominal power and 5% type I error under the Weibull model (two-sided test).
| R = m2/m1 | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
|
| |||||||||||
| κ | Test | 1.1 | 1.2 | 1.3 | 1.4 | 1.5 | 1.6 | 1.7 | 1.8 | 1.9 | 2.0 |
| 0.5 | Schoenfeld | 12335 (90) | 3406 (90) | 1662 (91) | 1020 (90) | 709 (90) | 533 (90) | 422 (90) | 347 (90) | 293 (90) | 253 (90) |
| Sprott | 12334 (90) | 3405 (90) | 1661 (90) | 1019 (90) | 708 (90) | 532 (90) | 421 (90) | 346 (91) | 292 (90) | 252 (90) | |
| Log-rank | 12333 (90) | 3405 (90) | 1660 (90) | 1019 (90) | 708 (90) | 531 (90) | 420 (90) | 345 (90) | 291 (90) | 251 (90) | |
|
| |||||||||||
| 1 | Schoenfeld | 2510 (90) | 693 (90) | 338 (90) | 208 (90) | 145 (90) | 109 (90) | 87 (91) | 71 (90) | 61 (91) | 53 (91) |
| Sprott | 2510 (90) | 693 (90) | 338 (90) | 208 (90) | 145 (90) | 109 (90) | 87 (91) | 72 (91) | 61 (91) | 53 (91) | |
| Log-rank | 2510 (90) | 693 (90) | 338 (90) | 208 (89) | 144 (90) | 109 (89) | 86 (89) | 71 (90) | 60 (90) | 52 (90) | |
|
| |||||||||||
| 2 | Schoenfeld | 582 (90) | 160 (90) | 78 (90) | 48 (90) | 33 (90) | 25 (90) | 20 (91) | 16 (90) | 14 (91) | 12 (91) |
| Sprott | 583 (90) | 161 (91) | 79 (91) | 49 (91) | 34 (91) | 26 (91) | 21 (92) | 17 (92) | 15 (93) | 13 (93) | |
| Log-rank | 582 (90) | 160 (89) | 78 (89) | 48 (90) | 33 (88) | 25 (88) | 20 (88) | 16 (86) | 14 (87) | 12 (87) | |
Empirical powers (in brackets) for the corresponding sample sizes were calculated based on 20,000 simulated runs under the same parameter setting.