Dear All,

does somebody have any suggestions with respect to the following problem?

I am running ANOVA with 3 independent variables (sex, grade and clustermembership). Because of unbalanced data (unequal cell sizes) and the probability of interaction between rows and colums, Type III Sums of squares are recommended, and I therefore apply the regression approach ( 'Method= Unique'). I have used the Scheffe post-hoc test to test for main effects, e.g. to explore the differences between different grades when ANOVA reports main effects for grade. It turns out that ANOVA sometimes gives sign. main effects, but Scheffe says 'no two groups are different at the .050 level'. Is Scheffe appropriate as a post-hoc test for main effects when the ANOVA regression approach is used? If not, what test is?

thanks in advance

Ron Scholte