Hello,
I am using 3dISC to model data from a mixed-effects design. I have 3 between-subject levels (patients with drug; patients with placebo; controls with placebo) and 4 within-subject levels (4 types of videos all subjects watched).
I am not interested in the between-group ISC and started by building a model disregarding the within-subject effects (averaging the pairwise correlation matrices from all videos using 3dCalc). Based on example 2 I came up with the following model:
3dISC -prefix ISC2b -jobs 12 \
-model '0+grp+(1|Subj1)+(1|Subj2)' \
-gltCode ave '0.333 0 0 0.333 0 0.333' \
-gltCode G11 '1 0 0 0 0 0' \
-gltCode G22 '0 0 0 1 0 0 ' \
-gltCode G33 '0 0 0 0 0 1' \
-gltCode G11vG22 '1 0 0 -1 0 0' \
-gltCode G11vG33 '1 0 0 0 0 -1' \
-gltCode G22vG33 '0 0 0 1 0 -1' \
-dataTable \
Subj1 Subj2 grp InputFile \
s1 s2 G11 s1_2+tlrc \
...
s1 s3 G12 s1_3+tlr \
...
s1 s4 G13 s1_4+tlr \
...
s3 s6 G22 s3_6+tlr \
...
s3 s4 G23 s3_4+tlr \
...
s4 s7 G33 s4_s7+tlr \
First I wanted to check if this looks correct or if the reverse group comparison constrasts need to be added:
-gltCode G11vG22 '-1 0 0 1 0 0' \
-gltCode G11vG33 '-1 0 0 0 0 1' \
-gltCode G22vG33 '0 0 0 -1 0 1' \
Secondly, I wanted to asked if the best way to model for the within subject conditions is as described in Example 5.
For example, from those 4 types of videos, 3 types are social and 1 type is non-social. I wanted to look at the interaction effect of "group by socialness" and at the effect of group within social videos only.
Is there a way to run a single model with all of these effects or do I have to run separate ISC models (i.e. a model with all videos, as above; a model only with social videos; and a model for the socialvs.non-social).
Thanks for any input in advance!
Cheers,
Vasco
