Hi,
> Only when dividing the ResMS map by the effective df's (SPM.xX.erdf) did I
> get a correct estimation of the individual SD,
> (for a particular contrast) using:
>
> SDmap = sqrt ( (resms_map/spm.xX.erdf) * (contrast'*SPM.xX.Bcov*contrast) );
>
>
> which is based on the formula for t :
>
> spmT(i) = contrast'*beta(i)/sqrt(ResMS * contrast' * SPM.xX.Bcov * contrast)
>
> Is one of the creators of SPM2 aware of this (minor) issue? Why do you call
> the images ResMS while it appears to be ResSS.
I think you just run into this bug:
http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0503&L=SPM&P=R31846&I=-3
As you see, it doesn't affect SPM in its use of the ResMS image,
because it uses a stored version of the vol struct, which has the
correct scaling...
Best,
Matthew
|