Dear R Gurus: I vaguely remember reading that if interaction was present in a factorial design, then the main effect results were suspect. However, I was reading a text which now uses the tests for main effects even if interaction is present. Which is correct, please? Thanks, Edna Bell