Table 5 Type I error rate when estimating under the null hypothesis of no difference between arms, with 30% missing data in same direction for each of the three simulation methodsa
k, clusters per arm | m, subjects per cluster | Method 1 | Method 2 | Method 3 | |||
|---|---|---|---|---|---|---|---|
ICC = 0.01 | ICC = 0.1 | ICC = 0.01 | ICC = 0.1 | ICC = 0.01 | ICC = 0.1 | ||
5 | 10 | 3.9 | 6.3 | 2.7 | 4.4 | 4.9 | 6.5 |
20 | 4.1 | 5.6 | 3.7 | 6.1 | 4.7 | 7.8 | |
50 | 5.2 | 4.4 | 5.4 | 4.1 | 4.8 | 6.8 | |
10 | 10 | 4.1 | 6.0 | 4.8 | 5.6 | 5.0 | 5.6 |
20 | 5.0 | 6.2 | 4.9 | 5.7 | 5.3 | 7.0 | |
50 | 4.7 | 4.2 | 5.3 | 5.4 | 6.4 | 8.1 | |
20 | 10 | 5.0 | 4.3 | 5.3 | 4.1 | 4.3 | 6.4 |
20 | 4.3 | 6.2 | 5.8 | 3.8 | 5.0 | 5.7 | |
50 | 4.8 | 5.2 | 4.8 | 4.3 | 5.6 | 5.7 | |
50 | 10 | 5.0 | 5.8 | 4.8 | 4.9 | 6.5 | 6.5 |
20 | 5.5 | 5.5 | 5.1 | 5.0 | 6.1 | 6.3 | |
50 | 5.2 | 4.7 | 4.5 | 5.1 | 6.4 | 6.7 | |