I'm going to run some analyses and have some doubts with respect to the general approach that I should follow. We want to evaluate differences in cortical thickness in three different groups. Shoud I perform an ANOVA (to know if there is a group effect) followed by pairwise comparisons (to localize the difference)? If so, what should I do if results from the ANOVA and pairwise comparisons are localized in different regions? Shoud I restrict the pairwise contrasts to regions where ANOVA revealed significant differences?