MFS, thanks for your suggestion. My thought *have* traveled that route, and I agree that a quad is more likely to have an (one) engine failure than a twin on pure statistics alone. However, since the gradients are calculated based on that we are already in OEI conditions, I figured that "safer" would mean "have a larger margin to disaster" and that 3/4 eng op would mean a larger margin than 1/2.
By the same statistical reasoning, you could argue that the probability of *another* engine failure of one of the remaining three quad engines is larger than failure of the remaining single twin engine, but I kind of thought that would be offset by the quad being more flyable on two remaining engines than the twin on none!
Still confused, not necessarily on a higher level...