Say I have a significant interaction for a 2x3 within-subjects ANOVA. If I have a theoretical justification for only running a single paired samples t-test to look at simple main effects (even when multiple t-tests are possible), do I still need to use a Bonferroni correction when doing so?
Sub-question: when looking at the main effect of a factor with 2 levels, should you get the same result as if you were doing a t-test between the 2 levels?
