04-04-2014 09:45 AM
Does it make sense to use PCA scores in factorial MANOVA, and then, using raw discriminant function coefficients (created by MANOVA) to calculate canonical discriminant
function scores (for each stat. signif. function/effect to maximally differentiate the groups). From this, doing factorial ANOVA on each signif. canon. discrim. function for
interacations, and then follow up signif. interaction with simple effect tests...