As BlueTrain says, the tests are a comparison of ammo, not real world results.
There is not and can not be testing for real world results. There are far too many variables. There would always be objections.
Simulate a 200 pound human and what happens if your attacker is 300 or 120?
How would you reliably replicate the possible angles of impact?
One persons bones being stronger or weaker or thicker or thinner?
Thickness of the down in a jacket? Different materials used as insulation? 3 layers of clothing or 4 or 2?
Over penetration is of no concern IMO. Hit rates in shooting are commonly in the 10, 20, 30% range. Even if it's 5 out of 6 (virtually unheard of), that 6th round is far more dangerous than the other 5.