Scientific American yesterday posted a nice summary of the research on effectiveness of psychotherapy.
From the article by Hal Arkowitz and Scott Lilienfeld:
In light of such findings, a search for a therapist should at least sometimes involve a consideration of the type of treatment he or she practices. It is true that ingredients, such as empathy, that cut across effective therapies are potent and that various established techniques are roughly equivalent for a broad range of difficulties. Yet under certain circumstances, the therapeutic method can matter. For example, if a clinician espouses an approach outside the scientific mainstream—one that does not fall under the broad categories we have listed here—you should not assume that this treatment will be as helpful as others.
One of the big puzzles of psychological treatment is why obviously unhelpful methods can lead clients to feel better. As noted in this article, some components of success cut across methods. A reparative therapist, for example, might explain a false set of facts to a client in an empathetic manner which might then provide a false sense of reassurance. Even though the therapy only minimally provides any effect on sexuality, the client may still feel better due to the working of the common factors.