Dear SPM users and developers,
I am performing a VBM in SPM8, but I got stuck when constructing contrasts for correlation.
I have 2 variables of interest A and B, and other variables to control for (c,d).
In a model including ONLY ONE of the variables of interest (A, c, d), the SPM of correlation were the expected by using the contrast [1].
However, if I include both variables of interest and control ones (A, B, c, d), the SPM resulting by using the contrast [1,0] is completely different than the reported with the former simpler model, and has no sense at all.
Since I believe that both approaches should be the same (or approximately), I am wondering in which step of my reasoning I am wrong.
Thank you in advance for your kindly comments.
|