*Sorry if this appears twice! I seem to be having problems with my email…
Hi everyone,
Does anyone know of a good text regarding higher order interactions? (or know how to interpret them?)
In some of my data, I have a significant four way interaction. Logically, I would test this by assessing whether a three way interaction differs on the level of of a fourth variable. For example, for variables A, B, C and D, I would do the following…
* Assess whether the interaction between B*C*D differs at each level of A
* Assess whether the interaction between A*C*D differs at each level of B
* Assess whether the interaction between A*B*D differs at each level of C
* Assess whether the interaction between A*B*C differs at each level of D
Firstly, I wanted to verify if this is correct. I was wondering whether I have to systematically go through every possible three way interaction (as above) or can I just look at the ANOVA model and test the three way interactions that are significant? I am pretty sure its the former, but thought I would check.
Secondly, I worry that because some of the variables have more than two levels, the number of comparisons would very likely result in a type 1 error and be meaningless to interpret anyway. If so, has anyone come across any papers who have overlooked the interaction for these reasons?
Any advice/good references would be greatly appreciated.
Cheers,
Kyle
|