3-way ANOVA with 4 conditions

AFNI Message Board
1

Hi CONN experts,

I want to look at a contrast I cannot quite wrap my head around in terms of implementing it in the GUI.

I have two Subject groups - Smokers and Controls - and four conditions - Repeated errors valued with 5cents i.e. EE5, the same valued with 50cents EE50 as well as corrected errors valued with 5cents EC5 and 50 cents EE50.

I want to look at an effect of smoking group (Smoker, Control) * magnitude (5 cents, 50 cents) * future performance (corrected, repeated) for connectivity between Hippocampus, Insula and ACC.

For the between subject contrast I would select Controls > Smokers and insert [1 -1]

Now since all the conditions are in one row, it is not as easy to distinguish. Can I write [1 1 -1 -1; 1 -1 1 -1] and would this give me a value int he directionality that corrected is stronger than repeated and 5 cent stronger than 50cents?

For the region I would select 1 as I have no prediction for directionality.

I have looked into this contrast for BOLD response and it may be that it doesnt make sense for functional connectivity. Any advise on implementation or how to read the result is much appreciated.

Cheers,
Leonie

2

Leonie,

I’m not so sure why you posted your CONN question on the AFNI message board. Such a group analysis can be performed through 3dMVM in case you’d like to try it out in AFNI.

For the between subject contrast I would select Controls > Smokers and insert [1 -1]

If univariate GLM approach is adopted in CONN, the test for group difference as shown above would be incorrect.

Can I write [1 1 -1 -1; 1 -1 1 -1] and would this give me a value int he directionality that corrected
is stronger than repeated and 5 cent stronger than 50cents?

It’s not clear to me what you want to achieve with the above F-test, but it’s worth mentioning that testing for the main effect of each factor (as well as the interaction of group with one of the two repeated-measure factors) is invalid under the univariate GLM platform.