Hi Edward,
In your design you already define groups using the three columns scz - rel - con. Randomise will use the ones according to the contrast you created defining groups. So the first contrast gets -1 in the first column and +1 in the last column, looking for Con>Scz area's. Can you explain a little more about the 7th contrast having Con > Rel,Scz? Is it correctly to set -1, -1 versus 2? And how do you interpret this output?
More information on your question regarding the warning in randomise can be found here:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=FSL;71dc5213.1005
The most important is to be sure about what you did, demeaning the columns. Also some people wrote about using the -D option, for demeaning your data if you are not modelling the mean in the design matrix.
Another comment is on the potential influence of confounders. Consider adding them in order to control for variables like age, gender, handedness and education. This may vary in groups. It is also known that patients with schizophrenia age differently from healthy controls (Jones (2006) in HBM 27).
Kind regards,
Stijn Michielse
Research Assistant
Dept. Psychiatry and Neuropsychology
Maastricht University
E-mail: [log in to unmask]
|