dear list,
we need a bit of advice about designing the correct factorial model in spm5.
our design is as follows:
main factor 1 = 2 levels (2 sessions as replication)
factor 2 = 3 levels
factor 3 = 2 levels
so we have 12 conditions in a 2 x 3 x 2 design. we want to test the
interaction effect between the 3 factors in a group of n subjects.
our problem is that in factor 2 there are no true 3 levels. the design was
planned with 3 conditions for factor 2, the first condition should serve as
a baseline contrast condition. so actually we would like to have only 2
levels in factor 2.
so now we are not sure, which design we should choose.
we would prefer to design a "simple" contrast model at the first step with a
column for each factor and level (=12 columns).
would it be appropriate to
a) model the interaction between the 3 factors in this design?
b) or to perform the "baseline" contrasts and transfer the resulting 8
con-images into a second factorial design (with a 2x2x2 design).
thanks a lot
markus
|