Hi,
> Two regressors will only fully correlate if their onsets are very close together... right??? this is not the case, I checked already...
> I have 12 runs with 7 event types... 5 of the event types carry a parametric modulator... a gray field sistematically appears when
> any one of this five conditions has less than 5 events within a run... this holds for my five subjects... it seems a bit more than a
> coincidence to me...
could it be, that this regressor could be expressed by a linear combination of the others? In that case, you would also expect, that the beta for this regressor is not uniquely specified, i.e. appears gray in the estimability.
So, to be more general, every time, when a regressor could be expressed by any linear combination of the other regressors in the design matrix, beta is not uniquely estimable. This includes all regressors, also those for the modulation and additional covariates, etc.!
Perhaps, you have the situation, that one regressor could be expressed by the others, when it has only 4 events?
Good luck,
Karsten
--
-------------------------------------------------------------
Karsten Specht, PhD
Department of Biological and Medical Psychology
University of Bergen
Jonas Lies vei 91
5009 Bergen
Norway
Tel.: +47-555-86279
Fax: +47-555-89872
Mobile: +47-93605508
[log in to unmask]
http://fmri.uib.no/
|