Could you entertain the idea that it might be supported for a small number of disorders in certain types of patients, and not for others? It really is not a black and white thing, so I don't understand why you keep coming back to the issue as if it is.I am 100% interested in any empirical citations you can provide. Seriously.
Either the “hypothesis” is supported or it’s not. If it’s not, we should do away with it. The convenience of the lie doesn’t make it true.
Role of Self-medication in the Development of Comorbid Anxiety and Substance Use Disorders
Mood Disorder Self-Medication article
ETA: Some of the better papers that I have read on this topic.