Question: Suppose in a two-way factorial study, the fitted interactions abij are both statistically detectable and large in a practical sense.
(a) Is it reasonable to think of a Factor A effect without regard to the level of Factor B? Why or why not?
(b) Is it reasonable to think of a Factor A effect for each level of Factor B separately? Why or why not?
(c) Can Factor A effects be judged to be essentially the same at each separate level of Factor B? Why or why not?