Hi All,

A quick question: Is there an easy way to write out or in some other way determine constrast coefficients for a three-way anova? I've a nice book from the SAS series about linear models but they show only a scheme for devising constrasts in 2-way models. In my case, I've a 3 X 2 X 5 design, with the last factor random; the factors are completely crossed. I'm interested in simple main effects for the first factor "against" the second. So, I'd like to write a constrast for A1 for the two levels of B, a contrast for A2 at the two levels of B, and the same for A3. What I've got programmed, and what seems to do the trick, is

contrast 'A1' B 1 -1 A*B 1 0 0 -1 0 0; contrast 'A2' B 1 -1 A*B 0 1 0 0 -1 0; contrast 'A3' B 1 -1 A*B 0 0 1 0 0 -1;

Now, the random factor is not of substantive interest, and it doesn't play a role in the hypotheses. From what I can get from the SAS manuals, I don't have to any provide any coefficients for it since it doesn't matter computationally to the contrasts, and the same is true for the interactions in which it is involved.

So, do the contrasts I've provided above do what I think they are doing?

Thanks in advance.

Joe