Dear SPM experts,
I am interested in testing the correlation of continuous anxiety scores with BOLD (beta values) in a contrast of interest. In order to do so I first performed a 1st level analysis of the desired contrast. In the 2nd level specification I uploaded .nii files of the relevant contrast and added a covariate of participants' anxiety scores. In "Interactions" I tried selecting both "none" and "with factor 1" – and this yielded a similar design matrix (see fig. attached) - so i'd like to know if this matters somehow.
But what i'm really not sure about is the contrast definition in this case. The left side of the design matrix is termed "mean (unique)" and the right side is defined as "anxiety (unique)" (see fig. attached), and I was wondering about the following options for contrast defintion and their meaning:
(a) Will defining a contrast of (0 1) result in the desired correlation?
(b) Does the contrast (1 0) represent the activation yielded by the contrast while controlling for anxiety levels?
(c) What does a (-1 1) contrast mean in this case?
Finally, I have a more general question about covariates in SPM. Prior to using SPM I used Brainvoyager, wherein a covariate analysis resulted in a correlation coefficient r value for each voxel, thereby representing a correlation between beta values of a specific contrast and individual differences in each voxel. Is this method equivalent somehow to the covariate method implemented in SPM?
Thanks,
Ofir
|