Hi Anderson,
I've compared the approaches we discussed on this thread (modelling IQ in three EVs versus one EV, plus in both cases modelling the group as three EVs, then looking at group average [1 1 1 0 0 0] vs [1 0 0 0]). Your suggestion seems to be more powerful in general, but one result has disappeared, but it's not just sub-threshold, so it's not in the raw z-stat image.
My understanding is that both approaches would still fundamentally look for those regions that correlate with IQ, collapsed across the different intervention groups, so I can't understand what is going on.
Please could you suggest ideas on how I can examine the data to understand this further?
Thanks.
Sam
(thread link https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1508&L=FSL&F=&S=&X=2C99BA99C8431EEE5E&Y=samlouiserogers1986%40gmail.com&P=56783 )
|