My problem looks like the following: I have a dataset on poverty, I have divided the variables into those on per capita income, per capita expenditure, and number of people in the family (each subgroup has at least 4 variables). I did a PCA for each subgroup and got 4,3,2 components respectively. I divided each component into 4 levels. Now I want to examine the effect of these components on poverty index using ANOVA. Can I enter as many as 9 dependent variables and their interaction into ANOVA? Is this a statistically valid approach? I have not found a maximum of 3 variables in ANOVA on the internet. Maybe I should consider all variables separately?