Skip to main content

Table 5 Well-approximated FWER under scenario 1 using different sample sizes among subpopulations

From: Practical basket design for binary outcomes with control of family-wise error rate

 

FWER

Sample size of (\(n_1\), \(n_2\), \(n_3\), \(n_4\))

Independent

BHM-weak

BHM-strong

(20, 20, 20, 5)

3.1%

3.8%

3.1%

(20, 20, 20, 1)

0.4%

7.1%

3.4%

(20, 20, 20, 30)

0.7%

3.2%

3.4%

(5, 5, 5, 40)

6.1%

4.6%

3.4%

(15, 10, 5, 1)

3.3%

7.9%

3.4%

  1. BHM Bayesian hierarchical models