I would think a basic coast down test of some sort would work. Find a flat spot, get the car up to 80mph and let it coast in N measure the time from 75mph to 50mph. Do it enough to get a good average and consistent numbers then remove the spoiler and do the same run again.
Some variation of this would probably be easy for most people to test. I don't think saying always remove it is better though. Remember when I did a coast down test with my rear spoiler a while ago and it was actually better with the spoiler on
It was a tiny difference but it was repeatable so it had to actually be helping.