Hi Jesper:
All is well my side of the pond, hope things are
looking good your end.
Thanks for your response to my query on second
level interactions.
----------------------------
The "easy" way would be to use a "PET-models-
Single subject: -> Conditions and
Covariates" analysis where you enter group as
your conditions and your
depression ratings as your covariate. Then
chose "Center around condition
means" and "Interaction with condition".
The "problem" with that approach is
the masking, which will not work so well on your
contrast images.
---------------------------
I did try that approach but faced the problem of
having to use global scaling even though that
had already been completed at the first level -
is that something to worry about in addition to
the masking problem you raise?
----------------------------------
The "slightly trickier" way would be to use
the "Multiple regression" option
in "Basic models". It really wouldn't be too
hard for you to create the
appropriate regressors to implement your model,
but you could also "cheat" a
little. Run the analysis I suggested above.
Disregard the results, EXCEPT
the SPMcfg.mat file which you load into matlab
while still running SPM. Now
do the "Basic models->Multiple regression" and
enter the columns of xX.X
(which was contained in SPMcfg.mat) as your
regressors. That way all the
"centering stuff" is automatically taken care of
for you and there is less
risk of mistakes.
-------------------------------
I like a good cheat and the above seems ideal.
Is there an easy way of getting matlab to spit
out the contents of xX.X..? Sorry for the stupid
question but I have a very rudimentary
understanding of matlab once I step out of SPM
world. :-)
Many thanks,
Stuart.
UPMC MR Research Center
PUH B-804
200 Lothrop Street
Pittsburgh
PA 15213
Sent by Medscape Mail: Free Portable E-mail for Professionals on the Move
http://www.medscape.com
|