Hi all

I am using conn for analysing resting-state fMRI data and I have a question about which connectivity measure to select.  Our analyses include evaluating predictors of individual variability in connectivity (e.g., age vs. connectivity) and test-retest effects (effects of active vs. placebo intervention).

There are four connectivity measures: correlation (bivariate), correlation (semipartial),regression (bivariate) and regression (multivariate).

My questions are:
-         What is the difference between these four measures? I am especially curious to know how regression (bivariate) vs. correlation (bivariate) differs.
-         My reading of resting-state fMRI studies suggests to me that most studies use bivariate correlation.  Is there an explanation for why this is preferable to the other options?
-
I am certain that regression (bivariate) and correlation (bivariate) will not yield the same result as I have found a significant difference (paired t-test) between the connectivity maps I get using these two measures.
I imagine there is a good reason for implementing one of the four options but neither the FAQ nor the manual elaborate on the difference between them or the appropriate application of them.

Thank you in advance!

Kind regards
Camilla Borgsted Larsen