Hi everyone,
I think I have a potential solution to this, which is to use 3dLME, which can export the residuals, for the group level analysis. Afterward, I use 3dFWHMx to estimate the smoothness of the data, and 3dclustsim to run the monte carlo simulations. Here are example commands that I am using for my data. I understand that this is potentially a somewhat new approach, and that there are differences between 3dLME and 3dMVM, so feedback is appreciated.
3dLME
3dLME \
-prefix ${resdir}/3dLME.${cond}.${load}.${dur}.${beam}.healthy_subs.${subnum}_subs \
-jobs ${jobs} \
-model "cond*load" \
-ranEff '~1' \
-SS_type 3 \
-resid ${resdir}/3dLME.${cond}.${load}.${dur}.${beam}.resid.healthy_subs.${subnum}_subs \
-dataTable \
#data table goes here
3dFWHMx
3dFWHMx \
-detrend \
-input 3dLME.thr.r.${dur}.resid.healthy_subs.${subnum}_subs+orig \
| tee 3dLME.${dur}.blur_estimate.acf.${subnum}_subs.1d
3dClustSim
3dClustSim \
-mask 3dLME.thr.r.${dur}.resid.healthy_subs.${subnum}_subs+orig \
-dxyz 2 2 2 \
-acf $acf \
-MEGA \
-both \
-prefix 3dLME.${dur}.3dclustsim.acf.${subnum}_subs
best,
-nick