Currently I am working on a way to distinguish Delirious patients from patients who still suffer from anesthesia and controls. I use the EEG and calculate relative powers in four frequency bands (Delta, Theta, Alpha and Beta) in three different regions (Anterior, Central and Posterior) of the head. I used a three group (Delirium, Anesthesia and control) repeated measures (as the three regions of the head will be related to each other) factorial Anova with the relative power in a frequency band in a specific region as the within subject factor. Now I get some significant results and would like to do post hoc analysis. For region and group it is not a problem, but I am mostly interested in the interaction between region and group (if e.g. the anterior region in relative theta power differs between delirious, anesthesia and control patients). If this interaction is significant: Am I allowed to do separate one-way ANOVA's+post hoc on the three regions to see where the difference in groups is? What is a correct approach to determine the difference in groups for a specific region?