I don't know of any positive evidence. I suppose I could dig up negative evidence without too much trouble but a lot of my conviction comes from more than 25 years of clinical experience. Looking at disorders such as major depression, the majority of anxiety disorders, bipolar disorder, and schizophrenia as examples, the predominance of research findings tell us quite clearly that the most effective and enduring treatments are those that combine medication and psychotherapy, with the psychotherapy component going beyond simple insight therapy.
If you think about it, there's no logical reason to expect otherwise: Insight tells you why you are feeling a certain way or why you are behaving a certian way -- it doesn't in itself tell you how to feel differently or how to behave differently. And when the disorder is at least partly biologically- or neurochemically-based, there is a need to treat that part of it: In the case of bipolar disorder, knowing that you suffer from severe mood swings and irrational thinking isn't going to help prevent either.
While I accept that insight-oriented therapies have some value, I don't see any evidence anywhere that they are sufficient as a treatment strategy. Look at the objective research into classical psychoanalysis: The course of treatment takes an enormous investment in time, up to 5 sesssions per week for several years, and even then there is little direct evidence that the process in itself helps much beyond the concentrated support and attention of the therapist over that period of time.