It turns out that the evidence for “empirically supported treatments” in mental health is murky at best; according to a meta-study of such ESTs (via MetaFilter): 20% “performed well”, 30% “had mixed results”, and 50% had “subpar outcomes”.