Dear List,
For an fMRI study we wonder which con.images one could best take to
the second level per subject.
1) One could take non-differential contrasts to the second level, i.e.
2 conditions + 2 task sessions:
1 0 1 0 and 0 1 0 1 = averaged per and over condition (two con.img for
1 subject). And on the second level test 1 -1 (within subject).
2) However one could also bring differential contrast to the second
level, i.e. 1 -1 1 -1 (one con.img).
For us the first option still makes more sense as a second level
analyses is meant to enable the generalisation to a broader
population. Using differential contrasts we would already remove
between subject and between group variance, i.e. we would not need to
specify a subject factor nor a group facor on the second level?
While with non-differential contrasts we would keep all the between
subject and between group variance (more variance is better for
statistics, right?). In this case we would need to model subjects in
the design matrix on the second level as specified in the tutorial
from Jan and Darren.
Please help me to figure out which approach is preferred.
If have overlooked any documents or emails please feel free to rub it
in as we have not been able to find an answer to this question
ourselves.
Thank you in advance and with kind regards,
Simone Reinders.
King's College London
Institute of Psychiatry (IoP)
De Crespigny Park
United Kingdom
|