Dear Ignacio,
> I have been getting some ugly gray fields in the uniquely specified betas bar, in the lower part of the design matrix...
> I usually get them whenever two events overlap, but this time I checked all my onsets and nothing was wrong... by trial error
> I discovered that you need to have at least 4-5 events in any condition for that field to become white again... can anyone
> enlighten me on this issue??? many thanks!!!
this always happens when two regressors are not independent, i.e. when they are fully correlated (negatively or positively). Therefore, you should check also the design orthogonality (use 'review design' for that) to figure out, which regressors are probably correlated (or mis-specified). Be aware, that this display of 'beta estimability' is not something like a 'green' light, i.e. which has to become white. It just says you, whether the regressors are independent our not. For example, you specified a block design with an explicitly model baseline condition. Than this grey fields are telling you, that only difference contrasts are allowed, i.e. task - baseline, since the modelled regressors are not independent in this case (either there is a task or a baseline).
Your experience with the 4-5 events per condition is interesting but just accidently and not a role of thumb (but, of course, you should generally have more than only 4 events for getting useful results)
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/
|