Table 1. Required sample sizes (N) per group in a group comparison design as a function of the number of fMRI volumes (influencing precision) and the expected size of the raw group difference in MSE.
| Amount of Data | Scale | MSE Precision | Estimated group-wise N required to achieve a specific raw group difference in MSE of: | ||||
|---|---|---|---|---|---|---|---|
| 0.05 | 0.10 | 0.15 | |||||
| Statistical Power (1−β) | 0.80 | 10 × 16 | 1 | 0.09 | 51 | 13 | 6 |
| 5 | 1.80 | >1000 | >1000 | >1000 | |||
| 10 × 32 | 1 | 0.08 | 41 | 11 | 5 | ||
| 5 | 0.55 | >1000 | 474 | 211 | |||
| 10 × 64 | 1 | 0.07 | 31 | 8 | 4 | ||
| 5 | 0.25 | 392 | 98 | 44 | |||
| 885 | 1 | 0.06 | 23 | 6 | 3 | ||
| 5 | 0.10 | 63 | 16 | 7 | |||
| 0.90 | 10 × 16 | 1 | 0.09 | 68 | 17 | 8 | |
| 5 | 1.80 | >1000 | >1000 | >1000 | |||
| 10 × 32 | 1 | 0.08 | 54 | 14 | 6 | ||
| 5 | 0.55 | >1000 | 635 | 282 | |||
| 10 × 64 | 1 | 0.07 | 42 | 11 | 5 | ||
| 5 | 0.25 | 525 | 132 | 59 | |||
| 885 | 1 | 0.06 | 31 | 8 | 4 | ||
| 5 | 0.10 | 84 | 21 | 10 | |||
| 0.95 | 10 × 16 | 1 | 0.09 | 85 | 22 | 10 | |
| 5 | 1.80 | >1000 | >1000 | >1000 | |||
| 10 × 32 | 1 | 0.08 | 67 | 17 | 8 | ||
| 5 | 0.55 | >1000 | 785 | 349 | |||
| 10 × 64 | 1 | 0.07 | 51 | 13 | 6 | ||
| 5 | 0.25 | 649 | 163 | 73 | |||
| 885 | 1 | 0.06 | 38 | 10 | 5 | ||
| 5 | 0.10 | 104 | 26 | 12 | |||
Note. Precision is defined as the SD of MSE values and can therefore directly be used to approximate the effect size of the expected effect in the statistical power calculation. Precision values shown here are the 95th percentiles of the precision value distributions across voxels, thus providing a rather conservative estimation of required sample sizes.