Dear Andrew,

I am using NBS to compare functional connectivity between two non-paired groups with unequal sample sizes. Let us say that those groups have 3 and 4 subjects. I have two nuisance variables to regress-out.
I have used the following design matrix where the 3rd and 4th columns contain the values of the respective nuisance variables.
1 0 0.9 10
1 0 0.3 12
1 0 0.1 15
0 1 0.5 6
0 1 0.7 9
0 1 0.2 1
0 1 0.5 10
I am using the following contrasts for unpaired t-test:
[ -1 1 0 0] and [1 -1 0 0].
Could you please tell whether my understanding is correct?
Does nbs.GLM.y contain the residuals after regressing out the nuisance variables ? How can I access those residuals in MATLAB ?
Thanking you
Mc Palin

Hi MC,

your understanding and proposed design matrix is correct.

Note that nbs.GLM.y does not contain the residuals. It simply contains the raw connectivity values.

To access the residuals you will need to explicitly run the GLM in Matlab. NBS does not provide the residuals.

Andrew

Originally posted by Mchandra pal:

Dear Andrew,

Thank you for the feedbacks.
I have two further queries.
1) I have two paired groups and two nuisance variables to regress out from the functional connectivity network matrices. We are interested to detect the differences in functional connectivity values between the two groups. Let us say we have 3 subjects in each group. I am using the following design matrices.
1 0 0 1 0.2 0.4
0 1 0 1 0.5 0.3
0 0 1 1 0.1 0.6
1 0 0 -1 0.8 0.3
0 1 0 -1 0.4 0.6
0 0 0 -1 0.7 0.1
and
1 0 0 -1 0.2 0.4
0 1 0 -1 0.5 0.3
0 0 1 -1 0.1 0.6
1 0 0 1 0.8 0.3
0 1 0 1 0.4 0.6
0 0 0 1 0.7 0.1
contrast in each case:
0 0 0 1 0 0
exchange block in each case:
1 2 3 1 2 3
Are those choices correct?

2) My second query is related to the brain behaviour correlation analysis using NBS. I have 1 group. I will regress out one nuisance variable from the functional connectivity matrices. Now, I want to check the positive and negative association of 2 cognitive variables with the residual functional connectivity values using F-test. My design matrix is:
1 0.2 0.3 0.1
1 0.4 0.6 0.9
1 0.4 0.7 0.5

where, 1st column is for global mean, 2nd column: nuisance variable, 3rd and 4th: cognitive variables and I am using the following F test contrasts: 0 0 1 0 and 0 0 0 1. Are these correct?

Why do we need to consider global mean?
Thanking you,

MC

Dear MC,

Yes - your design matrices look correct, although I think you are missing a 1 in the third column of the last row.

It is is important to include the global mean, unless you are specifically testing a hypothesis about the mean.

I suggest first testing the effect of no nuisance confounds and then adding in the confounds one by one. This way, you may be able to understand which confounds are most influential.

Andrew

Originally posted by Mchandra pal: