Ignore the protocol entirely Always assume that any correlation proves causation. Throw all your data into a spreadsheet programme and report—as significant—any relationship between anything and everything if it helps your case. If you measure enough, some things are bound to be positive just by sheer luck.