 |
 |
Dear FSL users, We understand that FLAME can deal with unbalanced designs and would like to know if that applies to our data. We have run a blocked design experiment with two conditions, Task (n=15) and Control (n=15), in two groups of subjects, Groups A and B. We want to compare the task effects (Task>Control) in the two groups in a third-level analysis. We have decided to exclude Task blocks with incorrect performance from the analysis, which leaves us with 5 Task blocks in Group A, and 10 in Group B. Will the fact that Group A has fewer observations confound our results, or will the variance estimation in FLAME allow for this difference? We appreciate any feedback on this. Many thanks in advance! Hanne