Dear SPM experts:
I have a somewhat unwieldly paradigm that I modeled using a 3x2 flexible factorial design (task; levels A,B,C; set size, 2, 4 items).
(A2, B2, C2, A4, B4, C4)
Is it proper to conduct contrasts within the flexible factorial model that do not include all 6 condition regressors, if there is an a priori theoretical reason to do so? For example, if I wanted to compare 2 specific levels of a task collapsing across set size: [1 -1 0 1 -1 0], [-1 1 0 -1 1 0]. A reviewer has concerns regarding the number of degrees of freedom not being appropriate, since there are 6 regressors but only 4 would be in play for this contrast.
Any insights would be greatly appreciated.
|