Consider just doing a whole bunch of t-tests

This amounts to saying "There is something significant going on if there is anything significant seen"

This is the method of \textbf{Least Significant Difference} and has a very large potential error rate