Hi Anderson,
So would the following run the F test to see if there is a difference in males across all three groups (Both greater or less than)?
palm -i bh.lgi.5mm.mgz -m bh.mask.mgz -d design.mat -o results -n 500 -approx tail -s bh.white bh.avg_area.5mm.mgz -logp -d design.mat -t t.con -f f.con -fonly
where ....
t.con is
/ContrastName1 HC_M vs GRP2_M
/ContrastName2 GRP1_M vs GRP2_M
/NumWaves 8
/NumPoints 2
/Matrix
1 0 0 0 -1 0 0 0
0 0 1 0 -1 0 0 0
f.con is
/ContrastName1 HC_M vs GRP2_M
/ContrastName2 GRP1_M vs GRP2_M
/NumWaves 2
/NumPoints 1
/Matrix
1 1
1) Would I need to add -twotail or specify in the contrasts both directions (A>B, A<B) with -corrcon for this to test both directions or does the -fonly option already test both directions?
2) For testing just females would I run this separately or in the same model if I am expecting similar vs different results?
3) For the Group1vsGroup2 comparison (within the larger context of the F test which looks across all three groups), I wanted to account for disease severity as an additional covariate. Since the group sizes are not the same for this analysis (excludes HC) would it be better just to separately run this comparison?
Thanks,
Ajay
|