Mental health isn't a strictly subjective experience though. You can measure days called-in-sick. You can measure suicide count.

Cause and effect isn't necessary to reduce harm. You with me? It's nice to have certainly, but it is not required.

Imagine that I am a friend of yours. I tell you that my child is depressed and the doctor suggests pill1, it makes my child more lethargic. We try pill2 and the depression is gone.

I tell you that I am happy with the results of pill2 and that while I am interested in how it works and potential long-term consequences, I'm happy staying on pill2.

Is this imagined parent strictly better than a parent that refuses to try any pill until convinced of its mechanism and success? I would say yes.

Observation is the workhorse of the scientific paradigm, not idealized experimentation. The quantity, and thus value, of Inductive evidence is greater than that of Deductive evidence.