Anecdote time: I know a doctor who is convinced that he is right on his opinion that marijuana causes depression, has near-100% gateway properties and causes long term brain damage even after short periods of use. The rub? He works at a psychological hospital that get a lot of patients with acute drug-induced psychoses or other drug-related issues. From his sample, of course he comes to his conclusions! I have tried to explain the concept of 'sample bias' several times to him but either he is incapable of such numeric thinking or just refuses to do so.