Hi - it sounds like you're getting very different results when
running within-subject cross-session modelling at second-level.
Probably you should be using fixed-effects analysis at second level
for this stage of the analysis, hopefully this will solve your problem?
Cheers.
On 21 Feb 2007, at 00:31, Tyler Lesh wrote:
> Hi all,
>
> I've recently been getting the "Flame Stage 2 has given an
> abnormally large
> difference to stage 1" error when using 3.2beta. I noticed this
> problem
> mentioned before on the list and went to those posts first.
> However, for my
> data, rather than getting 4-5 of these messages I am showing
> hundreds (I
> tried to attach a log, but it was 180K zipped and over 2MB unzipped)!
>
> The design of the task is such that each subject has two runs for two
> lower-level feats. These two runs are fed into a higher level for
> each
> subject. Based on previous posts, it seems that the cause of this
> error is
> higher than expected variance. First of all, although I can look
> at the
> varcope, I'm not entirely sure what to look for. Second of all,
> the sheer
> number of error messages suggests that something has gone
> drastically wrong!
>
> I have run 6 subjects so far and all are showing these errors. The
> quantity
> of errors varies across subjects and contrasts and ranges from
> about 7 up
> into the hundreds.
>
> Any help would be greatly appreciated.
>
> Thanks,
>
> Tyler
------------------------------------------------------------------------
---
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
------------------------------------------------------------------------
---
|