Dear Noelia,
1) The order of the covariates in the design matrix does not matter.
With the "multiple covariates" option, you provide the set of covariates
as a matrix R (one column per covariate) and you can also provide their
names with a cell array uninspiringly called "names". This seems to be
undocumented at the moment:
https://github.com/spm/spm12/blob/r7487/config/spm_run_factorial_design.m#L571-L573
2) You should probably be using dummy variables, made of 1 and 0 to
encode each level.
Best regards,
Guillaume.
On 28/11/2019 07:48, Noelia Martinez Molina wrote:
> Dear SPM experts,
>
> I´d like to set up a second level analysis controlling for multiple confounding regressors and I´d need some help with a couple of things:
>
> 1) Does the order in which I enter the covariates influence the results? I know that it does for parametric modulators since SPM uses spm_orth to orthogonalize the modulators. And yet another thing: if I create a .mat file with R containing one column for each nuisance regressor, can this be pass as 2 cell arrays, one for "names" and another for "values" of each regressor as we do for multiple conditions in the first level (names, onsets, durations)?
>
> 2) One of the nuisance regressors is a categorical variable, handedness, coded as 1,2,3. Should I use these values and then "No centering" in this case?
>
> Thank you very much in advance for considering my questions.
>
> Best wishes,
> Noelia
>
--
Guillaume Flandin, PhD
Wellcome Centre for Human Neuroimaging
UCL Queen Square Institute of Neurology
London WC1N 3BG
|