Dear Fabiana,
>
> Just a brief question. I have a fMRI design. I estimated the single
> subject design matrices doing the global normalization. Is it correct?
> If not, why? Could it affect the second-level analysis?
Yes it is. Global normalization at the first level will usually make
the second-level analysis much more sensitive because you are removing
a substantial source of spurious variance from the contrasts.
With best wishes - Karl
|