Hi Alexandre,
If the models (e.g. design matrix and settings) are the same, then it doesn’t matter what SPM calls it. The same model will always gib the same result.
If the models are different, then you will get different results.
In your case, it sounds like the models are different as you mention one model with a constant and a covariate another model with a constant and several covariates.
The order of covariates does not matter.
Best,
Donald
> On Jan 9, 2018, at 4:31 AM, Alexandre Obert <[log in to unmask]> wrote:
>
> Dear all,
>
> I know that this subject was addressed before but I can't find clear answer to my question.
> I've got the intuition that testing for an effect of a covariate in a one-sample t-test is not equivalent to test for its effect in a multiple regression with additional covariates.
> Some tests I made seem to confirm this but I can't explain it clearly.
>
> Is this link to orthogonalisation?
>
> Regards,
>
> Alexandre
|