Table 2.

Empirical power estimates for one-sided 0.05 level test of equal survival (H₀:F₁(t) = F₂(t)) against shorter/longer survival of group 1 versus group 2 (H₁: F₁(t) > F₂(t)/H₁:F₁(t) < F₂(t)), F_j(t) ∼ exp(λ_j), j = 1, 2, (λ₁, λ₂) = (0.08, 0.04)/(0.04, 0.08), c₁ and c₂ refer to the percentages of censored observations

n1	n2	γ1	γ2	c1† (%)	c2† (%)	Log-rank†	Perm†	Heinze†	IPT	IPT*	IPZ	IPZ*
6	6	0.00	0.00	10/28	27/10	0.280/0.277	0.254/0.252	0.248/0.247	0.256/0.257	0.257/0.256	0.242/0.252	0.254/0.252
6	6	0.04	0.04	37/55	54/35	0.213/0.210	0.197/0.196	0.198/0.198	0.221/0.203	0.176/0.181	0.192/0.185	0.163/0.185
30	30	0.00	0.00	10/27	27/10	0.772/0.773	0.762/0.765	0.758/0.759	0.786/0.766	0.773/0.768	0.758/0.772	0.729/0.760
30	30	0.04	0.04	36/55	55/36	0.615/0.616	0.605/0.608	0.603/0.605	0.640/0.600	0.608/0.609	0.578/0.556	0.548/0.540
6	30	0.00	0.00	10/27	27/10	0.459/0.348	0.355/0.397	0.336/0.396	0.336/0.437	0.350/0.393	0.343/0.399	0.350/0.384
6	30	0.04	0.04	36/54	55/36	0.366/0.242	0.282/0.291	0.278/0.292	0.270/0.312	0.261/0.280	0.264/0.286	0.258/0.279
3	120	0.00	0.00	10/28	28/10	0.407/0.221	0.223/0.309	0.222/0.306	0.215/0.312	0.223/0.307	0.211/0.317	0.230/0.286
3	120	0.04	0.04	37/54	55/36	0.342/0.137	0.196/0.223	0.193/0.204	0.180/0.231	0.190/0.191	0.159/0.227	0.190/0.237
6	6	0.00	0.04	10/27	54/36	—‡	—	0.223/0.211	0.241/0.208	0.210/0.211	0.221/0.196	0.196/0.194
30	30	0.00	0.04	10/27	55/36	—	—	0.676/0.654	0.697/0.659	0.690/0.638	0.653/0.619	0.650/0.637
6	30	0.00	0.04	10/27	55/36	—	—	0.329/0.351	0.340/0.392	0.332/0.341	0.339/0.348	0.323/0.351
6	30	0.04	0.00	37/55	28/10	—	—	0.277/0.313	0.261/0.319	0.278/0.313	0.271/0.280	0.293/0.286
3	120	0.00	0.04	10/27	55/36	—	—	0.223/0.298	0.225/0.311	0.211/0.305	0.236/0.296	0.212/0.296
3	120	0.04	0.00	36/55	28/10	—	—	0.177/0.198	0.183/0.223	0.179/0.226	0.186/0.229	0.190/0.245

^*imputed from the true underlying F, G₁, and G₂.

^‡“—” indicates settings where the corresponding tests are not expected to maintain the nominal Type I error rates.

^†Taken from Heinze and others (2003).