Re: clinical vs statistical validity

Greg Bail (cal4pooh@uclink4.berkeley.edu)
Wed, 15 Oct 1997 20:41:34 -0700


Rob Adelman wrote:
>
> Dear All,
>
> Bob Green suggested I turn this subject over to the group to get some more
> perspectives. Please read on and respond as you wish.
>
> "Well, the old committee keeps me turning in circles...
> Can you fill me in, or refer me to publications or people, dealing with the
> acceptable standards for the statistical testing of differences between
> grids. That is; while we may interpret the grid changes (taken from two
> administrations of the grid to the same person at different points in time)
> from a clinical perspective, how can we be certain of the validity of our
> interpretation, and how do we determine which changes are statistically and
> clinically significant. I have a committee member trained in traditional
> statistical theory who is having difficulty seeing the science in
> constructivism. Any thoughts?"
>
> Thanks,
>
> Rob Adelman
> Texas A&M-Commerce

Rob-
As a student new to the complexities of constructivism, I can sympathise
with your statistician friend. I was first exposed to constructivist
ideas in a family therapy class in my MSW curriculum at UC Berkeley and
still have much to learn. Nonetheless, I have some thoughts....

Are you using any parallel measures to your development of the grids?
Clearly, that would be the most traditional approach to validity, but
another approach might be to calculate internal validities to see if
those measures remain stable over time.

From a more theoretical angle, I am interested about whether you've
constructed grids to cover your own interpretive/therapeutic process as
a constructively oriented therapist. I do not mean to complicate
things, but I am interested in learning more about the mechanics of
research bases for the constructivist therapies. A comparison of your
grids for your own process vs. those of your clients might be
illuminating.

Thanks! greg

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%