# When do three-factor interactions or higher be assumed negligible?

I am referring to fractional factorial designs where some two-factor interactions are aliased with three factor-interactions, and three-factor interactions are aliased together.

To simplify the analysis and the de-aliasing process, I would like to assume that three-factor interactions are negligible. I do not want to interpret them since they do not have a real meaning when drawing conclusions from the experiment.

Is there is a way to justify the assumption of three-factor interactions being negligible? I do NOT have knowledge that certain three-factor interactions will be negligible before conducting the experiment.