Hi - good question, it would be nice, as you say, particularly for group
analyses, to be able to insert "dummy" EVs. We don't currently have the
facility for that - it's on our TODO list.
You can't insert empty (all zero) EVs as that breaks the model fitting.
However, I have just tested an easy hack to insert nearly-zero EVs
(effectively as confounds). I found that the results on the other EVs and
contrasts were almost unaffected, so this is probably an ok workaround:
setup dummy EVs (with whichever EV number you need them to have). If you
need the dummy EV to also have an associated temporal derivative (ie if
the EV it is "replacing" would have had a temporal derivative) then leave
that turned on. For the EV details: turn off convolution. Set shape to
Square. Set timings to off=0 on=[TR period - e.g. 3] stop_after=[TR
period]. Then you should end up with a single "1" at t=0 and 0 everywhere
else.
Hopefully this will work ok for you, thanks, Steve.
On Mon, 19 May 2003, Anda van Stegeren wrote:
> In a comprehensive study we defined 10 events, where subjective ratings in
> categories were linked to fMRI data.
> These 10 events are defined by 3-column textfiles. However: some subjects
> have one of the 10 events missing, for example because they didn't use one
> of the possible ratingcategories.
>
> Is there a possibility of defining an event as a MISSING VALUE ?
> It would greatly help to leave all the contrasts I defined as they are and
> not having to define for each individual a new design,(and will make it a
> lot easier with the Higher-level analysis later.)
> thanks,
> Anda van Stegeren
>
Stephen M. Smith MA DPhil CEng MIEE
Associate Director, FMRIB and Analysis Research Coordinator
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
|