I am skeptical because your testing seems like paint splatter on a wall. If you want to compare numbers and have it mean anything you need to do multiple run ABA testing in controlled driving circumstances with the same weather and everything. This means open highways, cruise control, no winds, similar temperates, decently long trips, and multiple trials. And I'm not even all that particular about it.
If you can do some of that, it might mean more. But Popular Mechanics just spent $1800 doing that and decided it didn't work.