Table 5.

Simulation results for designs without a control arm in the null scenario p₁ = ⋯ = p₅ = 0.20, for N=250. Each η_m = Pr(N_E1> N_Ek + m), the probability that the number of patients randomized to arm C is at least m larger than the number randomized to arm E_k. All values are per-arm.

Method	Pr(Select)	Pr(Stop)	n̄(95% CI)	η10, η20, η30
AR(1, 0)	0.01	0.19	50 (10, 137)	0.42, 0.35, 0.28
			Total	250 (250, 250)
$\text{AR}(\frac{1}{2},0)$	0.01	0.26	50 (10, 110)	0.41, 0.33, 0.26
			Total	249 (250, 250)
$\text{AR}(\frac{n}{2N},0)$	0.01	0.25	50 (10, 118)	0.41, 0.34, 0.27
			Total	249 (250, 250)
AR(1, 0.1)	0.01	0.24	50 (10, 128)	0.41, 0.34, 0.26
			Total	250 (250, 250)
ER	0.01	0.32	50 (10, 97)	0.32, 0.23, 0.2
			Total	248 (250, 250)