Hi Jess
> I have conducted an imaging study for my Masters research project. I have
> had some difficulty with the analysis and I’m now resorting to conducting a
> fixed effects analysis on my data however I do not know how to do this using
> the batch editor in SPM8-can anyone tell me how to go about conducting a
> fixed effects analysis at group level? (as far as I can tell at secondary
> level the default for SPM 8 is random effects which I need to override
> somehow?)
To do a fixed effects analysis you will just be using a 1st level
model...i.e., modelling timeseries data. Each subject is entered as a
separate session (or each session of each subject, if you have
multiple sessions/subject). So you'll have one big design matrix with
all images for all subjects, and lots of columns modelling all your
conditions for each subject.
One other thought...if the reason for going to a fixed effects model
is that you are not seeing significant results when correcting for
whole-brain significance in a random effects model, there may be other
ways to help you out. For example, if you have strong a priori
anatomical hypotheses, correcting for whole-brain significance may
really be overkill. If you can reduce your search volume
significantly, you may still be able to detect (subtle) effects in
your data with a random effects analysis. The advantage being then you
are entitled to make inferences about a larger population based on
your study.
Hope this helps! Good luck.
Jonathan
|