. Author manuscript; available in PMC: 2016 May 4.

Published in final edited form as: Stat Sci. 2015;30(2):199–215. doi: 10.1214/14-STS504

Table 7.

Comparison of different four-arm trial designs of size T = 80. F: Fisher; α: type-I error; 1 − β: power; p* : expected proportion of patients in the trial assigned to the best treatment; ENS: expected number of patient successes; UB: upper bound

	Crit. value	H₀ : p₀ = p_i = 0.3 for i = 1, …, 3			H₁ : p_k = 0.3 + 0.1 × k, k = 0, 1, 2, 3
	Crit. value	α	p* (s.e.)	ENS (s.e.)	(1 − β)	p* (s.e.)	ENS (s.e.)
FR	F	0.019	0.251 (0.04)	24.01 (4.07)	0.300	0.250 (0.04)	35.99 (4.41)
TS	F	0.013	0.250 (0.07)	24.01 (4.15)	0.246	0.338 (0.08)	38.34 (4.68)
UCB	F	0.011	0.252 (0.06)	24.00 (4.12)	0.218	0.362 (0.08)	38.84 (4.71)
RBI	F	0.018	0.250 (0.03)	23.97 (4.06)	0.295	0.268 (0.03)	36.52 (4.41)
RGI	F	0.017	0.250 (0.02)	24.07 (4.07)	0.298	0.265 (0.03)	36.45 (4.36)
CB	F_a	0.017	0.270 (0.30)	23.98 (4.08)	0.056	0.419 (0.38)	40.92 (6.89)
WI	F_a	0.015	0.258 (0.22)	23.00 (4.14)	0.101	0.537 (0.31)	42.65 (6.02)
GI	F_a	0.000	0.251 (0.13)	23.97 (4.11)	0.002	0.492 (0.21)	41.60 (5.44)
CG	F_a	0.015	0.253 (0.13)	24.04 (4.13)	0.349	0.393 (0.16)	38.29 (4.82)

UB				24.00 (0.00)		1	48.00 (0.00)