Use 3dMVM or 3dLME?

AFNI Message Board
1

Dear experts,

I would like to ask you for advice on whether to use 3dMVM or 3dLME for my analyses.
My data consists of 1 between subjects factor (group (2 levels)) and 2 within subject factors: task conditions (5 levels), and time bins (4 levels; estimated with a FIR model). There’s no missing data. I would like to test (per condition) whether the shape of the activity over time differs based on group, so whether there is a groupxtime interaction.
I have initially set this up in 3dMVM (see code below), but now I have read the following in the 3dMVM manual, and I am unsure this is the right approach: “For an F-test involving a within-subject variable, use -glfCode with 3dLME instead.” Does this warning also apply to my analysis? Would 3dLME be a better approach after all?

3dMVM -prefix cond1_groupxtime
-wsVars "ConditionTime"
-bsVars “Group”
-num_glf 2
-glfLabel 1 ‘cond1_group’ -glfCode 1 'Condition : 1cond1 Group : 1group1 -1group2’
-glfLabel 2 ‘cond1_grouptime’ -glfCode 1 'Condition : 1cond1 Group : 1group1 -1group2 Time : 1t1 -1t2 & 1t2 -1t3 & 1t3 -1t4’ \

2

time bins (4 levels; estimated with a FIR model)

Why are there only 4 bins to capture the HDR curve?

I have initially set this up in 3dMVM (see code below), but now I have read the following in the 3dMVM manual, and I am unsure this is the right approach

3dMVM is the way to go, but don’t use the GLF. Read the following paper as for how to set up the inferences:

Chen, G., Saad, Z.S., Adleman, N.E., Leibenluft, E., Cox, R.W. (2015). tecting the subtle shape differences in hemodynamic responses at the
group level. Front. Neurosci., 26 October 2015. http://dx.doi.org/10.3389/fnins.2015.00375