Table 3. Comparison of Bayesian posterior probability and Fisher exact test regarding local power and type I error.

			Probability of arm B as winner
	Difference	Arm B vs. A	Bayesian posterior probability	Fisher exact test (one-sided)
Power	20%	40% vs. 20%	4%	0.7%
	15%	40% vs. 25%	11%	2%
		35% vs. 20%	2%	0.2%
Type I error	0%	20% vs. 20%	0.01%	<0.01%