I’m mostly used to using AFNI group level analysis tools I started out with like the 3dANOVAs but realize the new abundance of tools available for mixed modeling options and am wondering if one of those would better fit my situation than what I have previously used.
Currently I have a completely within-subjects design with 2 within-subjects variables. Variable 1 has 3 levels and variable 2 has 2 levels.
Ultimately this will be a mixed design with 2 groups (between subjects), 3 stimuli (within subjects), and 2 runs (within subjects). I just do not have the second group yet.
Because this is just a 3x2 within subjects design for now I am using 3dANOVA2 type 4 for my data analysis but am worried about within-subject variability. I looked into 3dMEMA but the help says it does not handle well within-subjects variables with more than 2 levels although Gang’s post here indicates it can maybe handle 3dANOVA3 -type 5 models which are similar to what I’ve got going here, I thought.
I have also seen reference to 3dMVM and 3dLME. From the docs it seems like either may be ok for my situation. I’m just not sure which would be the best. Can someone point me in the right direction or explain where the programs are most optimal? I feel like I’m missing out on a lot by sticking with the 3dANOVA family.
Final question: I noticed a note about masking in one of the docs. I’m using an afni_proc.py outline for most of my preprocessing but I’m specifically using ANTs for alignment because there are some issues and ANTs does the best job. Before I align my data I mask it and do subsequent proc.py processing on the masked data. Is that a problem with these new programs?
A sincere thanks for any insight into these cool new techniques.