Dear Helmut,
Thank you very much for your answer and especially for the link.
Regards
Marsel
On 08-Apr-16 6:13 PM, MRI More wrote:
> Dear Marsel,
>
> Choose "none" for "Serial correlations" when setting up first-level, single-subject models, then you'll get ordinary least squares estimates. The rationale behind the default AR(1) setting is explained in http://www.fil.ion.ucl.ac.uk/spm/doc/books/hbf2/pdfs/Ch10.pdf page 7 onward (~ OLS on whitened data and design is equivalent to WLS).
>
> Best
>
> Helmut
>
|