Dear Marco

It depends how you coded the design matrix. E.g. if you had 1s for patients and -1 for controls, then a positive bar means more positive connectivity for patients, and a negative bars means more negative connectivity for patients. Note that the interpretation of the units of  ‘connectivity’ depends on the underlying DCM. In vanilla DCM for fMRI, the between-region connections are positive for excitation and negative for inhibition, whereas the inhibitory self-connections on each region are positive for more inhibition and negative for less inhibition.

In case of a between-region connection showing a positive effect for the group covariate (patients +1, controls -1), would this effect be compatible to either just one or both of the following scenarios:
1. Stronger excitatory effect for patients vs controls (e.g., connection strengths for patients = +0.9, for controls = +0.1)
2. Weaker inhibitory effect for patients vs controls (e.g., connection strengths for patients = -0.1, for controls = -0.9)

And, if both scenarios are possible, can they be somehow disambiguated?

All the normal rules for regression models apply here. So let’s say you have two columns (regressors) in your design matrix: a column of 1s and a column indicating the group membership (1s and -1s). Also, if you have an uneven number of patients and controls, I assume you have mean-centred your group regressor, i.e.

X(:,2) = X(:,2) – mean(X(:,2));

The mean-centring allows the first column to be interpreted as the mean. Then you have two orthogonal effects: the group mean and the difference from the group mean due to diagnosis, and you should report both. E.g. if the estimated mean is 0.3Hz, and the estimated effect of diagnosis is +0.1Hz, this tells you that patients have a connection strength of 0.3+0.1=0.4Hz and the controls have connection strength 0.3-0.1=0.2Hz.

Hope that helps,

Peter

Thank you and best wishes,
Marco

-- 
Marco Tettamanti, PhD
Associate Professor
CIMeC - Center for Mind/Brain Sciences
University of Trento
Via delle Regole 101
I-38123 Mattarello (TN), Italy
Phone +39-0461-283077
Email: [log in to unmask]
Skype: mtettamanti
http://scholar.google.it/citations?user=x4qQl4AAAAAJ
http://www.unitn.it/cimec