Post hoc test after 3dLME

AFNI Message Board
1

In our study we have 4(levels) groups and 2(levels) conditions so we looked at the main effect of group x condition.
Now we want to look at the post-hoc that is the group differences between any 2 groups, to do that we have extracted the beta values form the regions of group x conditionresults and performed ANOVA in SPSS.
Would this be a right way of showing the post-hocs or should we just do the post-hoc ttest in AFNI using 3dLME model of group x condition?

Jung

2

Would this be a right way of showing the post-hocs or should we just do the post-hoc ttest in AFNI using 3dLME model of group x condition?

It would be more preferable to obtain all the effects from one unified model. So 3dLME is the way to go for those post hoc tests.

3

Are these post hoc tests produced by 3dLME by itself or do we have to specify this?

For example I have group1 (condition1 condition2) group2 ( (condition1 condition2) group3 (condition1 condition2) so to look at the group difference between group1 and group2 for condition difference condition1 and condition2, I would have to do this

-gltLabel 1 group1_2 -gltcode 1 ‘group: 1group1 -1group2 Condition : 1condition1 -1condition2’

or 3dLME would auto generate this for you?

Jung

4

You have to specify all the post hoc tests. 3dLME and 3dMVM only provide those main effects and interactions.