Dear all, Is it dangerous to use step() during model simplification when I have an ANOVA design that is unbalanced (i.e. there is order dependence when entering the terms into the full model)? Thanks very much for your help! Kind regards, Christoph