Paul (and everyone else), it's not so much that I was being concilitory in my response. Rather, I was assuming that it goes without saying that if we have an empirical basis for expecting one therapy to be more effective than another, that should strongly influence our decision. We know that Cognitive Therapy and other cognitive-behavioral approaches have received considerable empirical support and that a number of traditional approaches have much less empirical support. Unfortunately, this doesn't give us the knowledge we would need to be able to choose the treatment approach which will work best for a given individual.
Gary is a good example of type of situation which clinicians frequently face. There is empirical support for the efficacy of CT as a treatment for Generalized Anxiety Disorder but a number of other treatment approaches have received empirical support (including other cognitive-behavioral approaches, biofeedback, progressive relaxation alone, psychodynamic therapy, and anxiolytic medication). I do not know of any treatment approaches for Paranoid Personality Disorder which have been empirically validated. So what treatment approach should I use? If I was to simply count the number of studies supporting each approach, I am not sure that CT would win.
I know from the empirical research (and personal experience) that CT is an appropriate treatment approach to use with Gary. However, we do not yet have enough of an empirical base to know if CT is generally superior to the other approaches to treating GAD (there are not enough comparative studies). We know even less about choosing the treatment approach which is best for a given individual.
I would love to be able to say that CT is superior to all the other treatment approaches, but that would be going way beyond the data. We can say that there is empirical evidence that it works well with a wide range of problems, and that is nothing to sneeze at.