The confusion ends when randomized studies are done. If the effects are hard to measure, very small, or present over decades, the confusion will then end when a

The confusion ends when randomized studies are done. If the effects are hard to measure, very small, or present over decades, the confusion will then end when a randomized study with a sufficiently large sample is done over decades. Until then, it's just trying to approximate that using techniques highly sensitive to noise.