Hi

On 3 Nov 2010, at 15:46, Kirstie Whitaker wrote:

Doh!  Yes - I had tried it with totally empty files and that didn't work, but a file which has 0 0 0 works fine.

I'm sorry to waste your time.....I probably could have figured that out all by myself......

But since I'm on a roll of asking questions, can I just check that even though I now have a rank deficient GLM, it's ok because I won't be testing any of the empty EVs?

Yes - hopefully it should work ok. If it runs through, then all is well.

Thank you again, so much (especially because your email was sent at 4am!?)

No problem - bad back so couldn't sleep……

Cheers.





Kx

On 2 November 2010 21:17, Stephen Smith <[log in to unmask]> wrote:
Hi - you should actually be able to select empty EVs in the GUI - have you tried that?
Cheers.


On 2 Nov 2010, at 20:55, Kirstie Whitaker wrote:

Dear Gurus,

I have an event related design for fMRI which has two types of trials (A and B).  Subjects can get these questions correct or incorrect, so now there are 4 possible EVs: Acor, Aincor, Bcor, Bincor.  Some subjects get a lot of questions wrong, but a large number get every question correct.  To make this even more complicated there are 3 different TYPES of error they can make: Ainc1 Ainc2 Ainc3.

(So if there was a particularly stupid subject they could have 8 different types of events: Acor, Bcor, Ainc1, Ainc2, Ainc3, Binc1, Binc2, Binc3)

Since I can't add an empty EV file into the design matrix I was wondering if if would be kosher to replace an empty EV text file (eg: Ainc1.txt if they made no A1 errors) with "0 0.001 0.001".

My understanding is that this will model an event right at the beginning of the scan, lasting 0.001 seconds and deweighted to be worth only 1/1000 of one of the regular events (which are real and actually happened).

I don't expect this EV to actually explain any variance but I'm hoping that it will allow me to compare subjects at the group level because now they will all have the same numbers of PEs and COPEs.

Does anyone have any other suggestions if this isn't a viable solution?

Thank you!

Kx

--
Kirstie Whitaker
Doctoral Candidate
Cognitive Control and Development Laboratory
Helen Wills Neuroscience Institute
University of California at Berkeley
tel: 510 684 2456
web: bungelab.berkeley.edu
blog: http://bungelab.blogspot.com/


---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------






--
Kirstie Whitaker
Doctoral Candidate
Cognitive Control and Development Laboratory
Helen Wills Neuroscience Institute
University of California at Berkeley
tel: 510 684 2456
web: bungelab.berkeley.edu
blog: http://bungelab.blogspot.com/


---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director,  Oxford University FMRIB Centre

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 222726  (fax 222717)
[log in to unmask]    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------