Dear Jorge,
> I have some problems understanding why I get always different beta
> parameters of the same subject data (same coordinate) looked in four
> different ways (SPM99b):
>
> a)Single subject analysis (just one subject in my design matrix). I
> display the corresponding beta_1.img file, go to an arbitrary
> coordinate and look at the beta (image intensity?) value.
>
> b)Single subject analysis. Results -> go to same coordinate as above ->
> plot contrast of parameter estimates -> effects of interest. The value
> I get in the graph (the same as beta(1,1) in the matlab terminal) is
> different than the value I obtained in a).
They should be the same. beta is simply the image values - it is not
changed in spm_graph
beta(i) = spm_sample_vol(xSDM.Vbeta(i),rcp(1),rcp(2),rcp(3),0);
> c) Fixed effects group analysis. In this case the data of my subject is
> together with that of other subjects. Other than that the analysis is
> the same as with the single subject analysis. I display the beta_##.img
> of the subject I am interested in, go to the same coordinate as above
> and get a beta value different from a) and b). Actually the whole beta
> image looks completely different, as if it was that corresponding to a
> different subject. But I checked that I am picking the correct
> beta_##.img.
This could be explained if the design matrix was not strictly subject-
separable (i.e. if you included any covariate - such as a global
estimator that spanned more than one subject).
> d) Fixed effects group analysis. Results -> go to the same coordinate
> as above -> plot contrast of parameter estimates -> effects of
> interest. The beta(##,1) value is different than a), b) and c).
>
> I was expecting to obtain the same beta values in these four ways of
> looking at the same subject data. Any ideas of why this should not be
> the case?
No I am sorry. You are right that (assuming a subject-separable design
matrix) all four values should be the same. I can only suggest you
check that the scalefactor and offset for the beta_??.img are 1 and 0
respectively and double check you are looking at the same voxels (the
cursor jumps to the nearest voxel in Y.mad in spm_graph which may not
be the one you selected from beta_??.img).
With best wishes - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|