Skip to main content
. 2008 Nov 16;10(2):310–323. doi: 10.1093/biostatistics/kxn037

Table 1.

Empirical probabilities of rejecting H0 in simulated trials with 2 interim reviews (at times T1 and T2) and the final analysis (at time T3)

Scenario B
 Scenario C
Procedure T1 T2 T3 Overall T1 T2 T3 Overall
MRCB 0.139 0.559 0.222 0.920 0.679 0.259 0.038 0.976
GST16 0.120 0.432 0.131 0.683 0.088 0.311 0.142 0.541
GST24 0.142 0.668 0.160 0.970 0.008 0.018 0.026 0.052
WSL1 0.163 0.498 0.187 0.848 0.340 0.211 0.061 0.612
WSL2 0.699 0.298 0.003 1.00 0.190 0.015 0.053 0.258
WSL3 0.284 0.709 0.007 1.00 0.143 0.082 0.078 0.303

MRCB: multivariate repeated confidence bands approach with weight An(·;Tk) and the max-norm. GST16 and GST24: group sequential procedures based on Hotelling's T2 test on response at weeks 16 and 24. WSL1, WSL2, and WSL3: extensions of the approach of Wei and others with linear means, using the identity, the true, or the pooled sample covariance matrices.