It seems to me that this discussion is not about the double-blind nature of
the experiment but rather about the appropriateness of the control
condition.
Let's assume that we use Stephen's design with two additional features 1) we
have several therapists from each school and 2) they are equally
enthusiastic about the efficacy of their treatments. Naturally, we will not
tell them about our hypothesis, they will assume that their therapy is the
more effective. Remember that some other dispassionate and blind observer
measures outcome. The study is now a true double-blind (Did I miss
something?), but the control condition allows for more than one
interpretation of any differences observed. Any difference could be due to
efficacy of the therapies or could be due to something else about
individuals who choose or study those therapies. For example, emotionally
warm therapists are generally more effective than cold regardless of therapy
used. (The temptation to pun in that sentence is almost irresistible.) I can
find a reference for this if anyone is interested.
I hate to open this can of worms again - From the consumers' view the above
design is extremely useful because it reflects the effectiveness of therapy
as they are likely to encounter it (high ecological validity). From a
scientific view (internal validity) the above design is still very useful,
but leaves unanswered questions about some confounds. As Paul has said, we
know more from this design than from the case study design (which is much
lower in both ecological and internal validity than this one). We still have
questions to answer, but they can be addressed in the next study in a series
or by some other researcher.
Dennis
Dennis M. Goff
Dept. of Psychology
Randolph-Macon Woman's College
Lynchburg, VA 24503
