 |
 |
Hi FSL experts, To clarify, in AFNI the results of the lower-level model (from 3ddeconvolve) of the single-subject are stats files. I believe these are equivalent to the "tstats" files in FSL that are generated from FEAT lower-level analysis. In AFNI, you use the t-stats files for the contrast of interest for each subject as the input for group-level models in 3dttest or 3dANOVA, ect. In FSL, you use the cope files. I believe this is because FSL is capable of doing mixed-effects modeling in flame, and therefore it uses the parameter estimates of the contrasts from the single-subject glm, instead of stats files. Is that a correct interpretation? Is there any scenario where using the cope images, vs. the statistics images, as the input for group-level would yield inconsistent results? I'm assuming there could be differences in magnitude, but one would not expect differences in the overall direction of the effects? Thanks, Michelle