afni.nimh.nih.gov

My 2 cents: A 3dANOVA4 might open a Pandora's box. 3-way interactions are already difficult to explain, let alone 4- or 5-way interactions. If ANOVA is the statistic of choice, I would highly recommend simplification of the research design to include only 3, rather than 4+ factors. This may not always be possible of course. If so, one should really consider regression analysis for a design that includes more than 3 variables.