Even with a high signal, you can have non-etiologically-based treatments. I'm sure with a high enough sample size, you could statistically conclude that giving

Even with a high signal, you can have non-etiologically-based treatments. I'm sure with a high enough sample size, you could statistically conclude that giving depressed people $1000/week their levels of depression as measured by the Beck Depression Inventory. That doesn't necessarily mean that being poor is etiologically connected to depression. I'm certain people were depressed before the invention of money. There is, however, a strong bias toward unnecessarily searching for a causal connection between treatment and disease, but also dismissing effective treatments where cause and treatment aren't sufficient connected in people's minds. eg: you're only treating the symptoms as if that can't substantially improve peoples' lives.