There are real drawbacks to basing supervision on the supervisee's description of a therapy session. The supervisor only hears about the things the supervisee noticed, remembered, and is willing to report. Obviously, this could mean that the supervisor misses out on important information. The supervisor also has limited opportunity to independently evaluate the supervisee's conceptualization, treatment plan, and interventions because all the data is filtered through the supervisee.

An apprenticeship model where the supervisee sits in and observes sessions the supervisor is conducting and where the supervisor sits in on sessions the supervisee is conducting, is not a bad concept (I think Dave Burns wrote a chapter on supervision based on this model). Unfortunately, in real life this would take a lot of time (and thus be expensive) and would be a real scheduling headache in most settings. Also, having a third party sitting in the room can really distort the therapist-client interaction.

There are other solutions that are more practical. Often the easiest way for the supervisor to model the desired behavior and to observe the supervisee's behavior is to review recordings of sessions. Supervisor and supervisee can review and discuss videos (or audio recordings) of the supervisor or other experienced therapists doing CT and can review recordings of the supervisee's sessions. This provides more detailed (and less biased) data that the supervisee's description of the session.

One can also observe live sessions through one-way mirrors or closed-circuit TV but this ends up being more time-consuming and more of a scheduling problem that reviewing tapes. An advantage of observing live sessions is that the supervisor can potentially coach the supervisee during the session. I've seen this done through the use of a "bug-in-the-ear" (i.e. an earpiece that relays the supervisor's instructions) or by phoning the supervisee during the session. I've also seen it done by having the therapist stop half-way through the session to step out of the room and consult with the supervisor. From what I've seen, each of these methods seems somewhat disruptive and I'm not sure they add much.

I do find that reviewing videos adds a lot.