Gang,

Hello, I've run with this dataset and ran into an issue where calculations in R and SPSS didn't match.

I ran my 2 x 2 mixed model using 3dMVM

Here is the breakdown:


2 Factors:

group - between subject 2 levels (with, without)
scan - within subject 2 levels (pre, post)

We generated a group x scan interaction map and looked at the interaction post-hoc comparisons from that map.

I was able to replicate the findings in R from extracted cluster peak voxels.

I get the same results in AFNI and R (using lme) for the following contrasts:

Group main effect - ( t contrast group1 vs group2, collapsed across scan)
Scan main effect - (t contrast pre vs post, collapsed across group)

group1_Scan1_vs_Scan2
group2_Scan1_vs_Scan2

However, for the following two contrasts:

scan1_Group1_vs_Group2
scan2_Group1_vs_Group2

I get different t scores in AFNI than I do R and SPSS. Not sure as to why. I thought maybe it was my model setup with lme, but I was able to replicate different findings in SPSS. Your help would greatly be appreciated.

Thank you for your time.

Michael