Dear expert, 
we have used CONN for our resting state fMRI and anxiety relationship paper, and got the reviews back asking to clarify the 2nd level analysis and also to compute effect size for reported F stats.

It's a 2 seeds (left, right amygdala) FC maps per subjects enter into the 2nd level model, with motion gender as covariate of not interests, and with anxiety as covariate of interest.  And we did an "OR" contrast, so eyes(2) for seed/source, and [0 0 1] for between subjects variable with anxiety as 1.

Can we say what's truly happening in 2nd level is a GLM as following?
Both left and right amygdala were entered into one combined GLM model in CONN. It's an F-test using an 'eye(2)' contrast which tests an OR conjunction of the left [1 0 ] and right [0 1] amygdala contrasts, to identify areas where the connectivity with any of the seed regions (left or right) is correlated with anxiety.


Also they asks to report the Effect Size from the F test, we obtained F(2,150)=14.5 from the CONN 2nd level GUI, is there a way to do offline analysis to obtain the ROI's effect size towards anxiety ?

I tried to extract ROI value from both left and right amygdala separately, and then do a FC~seed +anxiety + motion + gender using mixed linear model, but the F value here is totally different and with totally different df.  so I assume CONN computed the F stats very differently? 

Can you explain how it's computed in CONN the F stats?

Thank you!!

Xiaozhen You