Dear SPMer's:
Our question concerns creating mean images for subsequent random effects
analysis. Our study design is as follows:
RUN1: A B A B A B A B
RUN2: C B'C B'C B'C B'
RUN3: A B A B A B A B
RUN4: C B'C B'C B'C B'
A and C pertain to active conditions and B and B' are the control
conditions. The order of runs is counter-balanced across subjects. Now we
would like to create a mean image for each condition. Now we know that we
must collapse over the replications in order to create a single mean image
for each condition. Our question concerns which type of adjustment to use
for each mean image (scaling or ANCOVA). In the SPM96 help section on
adjusted means Andrew Holmes mentions that "...multiple runs *must* use the
same GM value, and should scale Grand mean *by subject*". We're not quite
sure what he means, and how does this apply to us? Thanks for your help.
Walter S. Marcantoni
Irena O'Brien
Université du Québec à Montréal
Laboratoire de Neuroscience de la Cognition
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|