 |
 |
I am trying to run a 3x3 independent measures ANOVA for TBSS (randomise). I have 3 levels, each with 3 factors. I have tried to follow the guide but have found examples for only a 2 way ANOVA with 3 levels, or a 3 way anova with 2 levels per factor. In the guide, there is a cell means approach, and a factor effects approach to constructing the design matrix of a 2 way ANOVA with one of the factors having 3 levels. Why would you choose one or the other of these approaches (are they equivalent)? Furthermore, how do you get an F-test for a main effect of the factor with 3 levels (since one has to either split the EV into 2, in the factor effects approach, or split the F-test into 2 T-contrasts in the cell means)? My question applies to both designing the matrix and the contrasts. Also, I would like to run an ANOVA with 3 factors, one of which has 3 levels, 2 of which have 2 levels. What example in the guide should I follow?