Hi Debby,
Assuming it is compatible with your task design, why not try a parametric modulation analysis at the first level and take the mean effect of stimulation for the second level analysis? That way, variations in intensity will have already been accounted for without requiring 4 separate covariates to eat up degrees of freedom at the second level. And, I think you can more easily assess whether there is actually an effect of intensity via the parametric regressor.
Best,
Tasha
________________________________________
From: SPM (Statistical Parametric Mapping) <[log in to unmask]> on behalf of Debby <[log in to unmask]>
Sent: Tuesday, June 20, 2023 5:00 PM
To: [log in to unmask]
Subject: [SPM] How are covariates handled in second level analyses?
Dear spm-users,
I am trying to investigate potential changes induced by an intervention (noninvasive brain stimulation) on rs-fMRI. So I am running a second level analysis based on a paired t-test with the inputs from my first-level analyses. So far so good. However, I would also like to correct for variations in the strength of the stimulation (4 scalar vectors for every subject). If I make a matrix for these covariate and put that in the 'multiple covariates' input in spm, I can indeed see that these are incorporated in the design matrix (see design here: https://www.dropbox.com/s/4fmvdgfh1sjeh5a/Schermafbeelding%202023-06-20%20om%2016.52.43.png?dl=0). But when I look into my results, I do not see any differences to my analysis without covariates so clearly I am doing something wrong.
If anyone has any advice on how to solve this issue, it would be highly appreciated.
Thank you so much.
Best,
Debby
|