The group average can lose weight without every member having lost weight.

The study compared "everyone who did this intervention" vs " everyone who did this other intervention". That seems perfectly fair to me, it's just answering a different question to your proposed comparison.