Dear Will,
thank you very much for your quick reply to my email. Can I ask you
a follow-up question?
The background of my question was, that I quite often use an F-contrast
that spans only the regressors if real interest (specified via 'columns
for reduced design') and use it to plot parameter estimates for these
regressors at a particular voxel (using the Plotting Util spm_graph).
However, as it turns out that the contrast weights of the F-contrast are
also scaled (and not equal to 1) then I do not obtain the true contrast
estimate from the con-images, but rather a somewhat "distorted" (scaled)
value and the plot does not reflect the underlying data.
Can you comment on this? Is it still OK to use the Plotting Util with
these kinds of F-contrast? How do you think about using these plots in
publications? Or would it be more appropriate to extract the data
directly from the beta-image (e.g. with spm_get_data), which would be
equivalent to SPM.xCon.c'*beta if the contrast weight matrix contains
only 1's and 0's, and the proceed with plotting e.g. the mean parameter
estimate for the group?
Thanks for your insights,
Jan
On 2005-01-17 (Mon) at 17:57:19 +0000, Will Penny <[log in to unmask]> wrote:
> Dear Jan,
>
> Your hunch is correct. These strange looking contrasts that are
> created automatically look strange because they have been operated upon
> by SPMs estimate of the error covariance (nonsphericity).
>
> I would ignore them and specify my own contrasts.
>
> An identity matrix that spans the whole design matrix may not be valid
> as eg. the column of 1's may be perfectly collinear with combinations of
> other columns.
>
> You could instead specify different contrasts that span the variables of
> interest eg. the F-contrast
>
> [1 -1 0; 0 1 -1]
>
> for a 1 x 3 ANOVA.
>
> Best,
>
> Will.
>
> Jan Gläscher wrote:
> >Dear list,
> >
> >I posted this message about a week ago, but got no answer. I wonder if
> >any of the SPM wizards is willing to comment on this issue.
> >
> >Thanks a lot.
> >Jan
> >
> >
> >----- Forwarded message from Jan Gläscher <[log in to unmask]>
> >-----
> >
> >From: Jan Gläscher <[log in to unmask]>
> >To: [log in to unmask]
> >Subject: confused about effects of interest contrast weights
> >
> >Dear Will, Karl and others,
> >
> >I have encountered a contrast weight matrix in the standard 'effects of
> >interest' F-contrast at the 2nd level (different ANOVAs) that I cannot
> >explain.
> >
> >I had expected an identity matrix as contrast weights, however the
> >actual values in SPM.xCon(1).c are actually quite dissimilar. In
> >addition, the same scaling effect of the contrasts weights occurs when
> >I specify an F-contrast via 'columns for reduced design' in the contrast
> >manager. In fact, when I try so specify an identity matrix
> >[eye(size(SPM.xX.X,2))] then this is not a valid contrast.
> >
> >Could somebody explain this scaling of contrast weights to me? Does it
> >have to do with the non-sphericity correction?
> >
> >Thanks a lot for your insights,
> >Jan
> >
> >
>
> --
> William D. Penny
> Wellcome Department of Imaging Neuroscience
> University College London
> 12 Queen Square
> London WC1N 3BG
>
> Tel: 020 7833 7475
> FAX: 020 7813 1420
> Email: [log in to unmask]
> URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
>
--
Jan Gläscher Neuroimage Nord
+49-40-42803-7890 (office) Dept. of Neurology, Bldg S10
+49-40-42803-9955 (fax) University Hospital Hamburg-Eppendorf
[log in to unmask] Martinistr. 52
20246 Hamburg
Germany
http://www.uke.uni-hamburg.de/zentren/neuro/neurologie/mitarbeiter/glaescher_jan.html
--------------------------------------------------------------------------------------
GnuPG/PGP key id: FEC4B55C
fingerprint: 5A36 1EF6 8472 117E 805A F240 3146 A410 FEC4 B55C
--------------------------------------------------------------------------------------
|