I am trying to set up few contrasts statements for a design with three different treatments (T, P, and K). T has three levels and P and K has four levels each. I do know that interaction T x K is significant. Now I want to test if response of K is linear for dependent variable Y. Then the next question is if this linear response holds true for each level of T.
Can somebody check if the following contrasts are right to answer second question?
Contrast "K linear in T1" T*K -3 -1 1 3;
Contrast "K linear in T2" T*K 0 0 0 0 -3 -1 1 3;
Contrast "K linear in T3" T*K 0 0 0 0 0 0 0 0 -3 -1 1 3;
I don’t know what should be the contrast for answering first question. While righting contrasts does it matter if out levels are equally spaced or not.
The variable T is categorical while P and K are unequally spaced quantitative variables.