I would like to use FEAT (ver. 5.98, CentOS 5) to compare brain activity
(within a single run/series) during correct and incorrect trials of a visual
task, for a single group. I'd then like to run a second-level mixed effects
analysis to look at population-level accuracy effects. For all subject, the
# of incorrect trials within a run is always less than the number of correct
trials - i.e., an unbalanced design. My assumption is that the height of
the parameter estimates for correct and incorrect should be the same,
regardless of the unequal trial numbers. I would then contrast the
unbalanced PEs to get a COPE map, still at the first level, and use this
COPE for second-level (group-wide) analyses. I have read some FSL listserv
posts (listed below) that seem to indicate that this combination of
unbalanced FILM (1st level) designs and FLAME (2nd level) should ultimately
be robust against the 1st-level imbalance. But I was wondering:
1) Could someone confirm this? I.e., is it legitimate to run analyses with
unequal #s of trials through FEAT/FILM in this way?
2) Legitimacy aside, are there any important consequences to doing this?
For example, would one predict that the condition with more trials would
yield more stable/significant clusters of activity?
3) Do you know of any citations from your group, or even from statistical
texts on which FILM might be based, that I could reference in a publication
to establish that it's valid to do this with FILM/FLAME?
Thanks,
John
Here are some listserv posts that touch on the issue:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0802&L=FSL&D=0&P=29294
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind05&L=FSL&P=R98896
https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind03&L=FSL&P=R14661
|