Counselor Constructs
Tue, 8 Aug 1995 20:45:28 -0400

Thanks to all of you who have responded to my dissertation problems. I
appreciate the suggestions offered and the questions you have posed. One
of the problems I have is that no one at my university ( at least not in my
department) is really tuned into PCP so I appreciate the opportunity to be
a part of this discussion list.

Jim asks:
>Have you tried to use any of the grid programs? Do you have any of the
>programs that have been developed?
> If you have not worked directly with grids, you might find that
>concretizing some of your questions might be of value. I will assume that
>have not already followed the course of action I will suggest!!!???

The only grid program I have tried was a copy of Omnigrid given to me by
Jack Adams- Webber. I did play around with it a bit, and I also read quite a
lot of information about other grid programs and the information one can
obtain using them. ( I was confused by a lot of what I read.) I'd appreciate
suggestions as to possible programs that I could use. However, I'm not sure
if a rep grid is the best way to approach the problem. Maybe if I explain
more about what I have tried, someone can tell me if there is a grid
program that will work.

I did create a set of role categories of clients and aquaintances and
used triads to elicit constructs from some of the master's level counseling
students, doctoral students and a couple of my professors , 10 people
altogether including myself. Then I tried to classify the constructs in
some meaningful way which is why I got interested in Goldberg's taxonomy. I
did come up with a classification system that seemed to fit the constructs.
My problem with using the triadic method to elicit constructs and creating a
grid and, more important, my advisor's problem with it brought me to the
same question Malcom posed.

>I am wondering if you are more interested in the content of constructs
>(what do counsellors understand of their clients?) or the structure of their
>construct systems (how do counsellors understand their clients?).

The constructs that were elicited from the triads were of basically four
types: Most were rather abstract constructs that fit into one of Costa and
McCrae's big five factors or one of the 16 factors from the 16PF. A few were
concrete behavioral constructs ex. "makes eye contact- doesn't make eye
contact" and some were clinical constructs ex. " borderline personality
disorder - no personality disorder" , then there were also a few that had to
do with the counselor's reaction to the individual. ex. " feel comfortable
with him- makes me nervous". My advisor also felt that since each subject
would fill in an individual from his/her group of clients and aquaintances
for each role category it would be impossible to draw any conclusions from
the data. I tried to explain that by eliciting constructs one is learning
about the counselor's system of constructs and that all clients will be
viewed by the counselor using the same system. He felt that all S's must
respond to the same individual or at least the same stimulus. He suggested
we provide the role category and have each subject list all words that they
could think of that would describe people that would fit in that category.
My reaction to this suggestion was that what we would get was the abstract
constructs like the "big five factors". It seems to me that counselors
differ from non- counselors in that they have more subbordinate constructs.
They have a greater degree of differentiation at the lower levels in the
hiarchy and more complex associations between these lower level constructs
and the higher level ones. At that point, I was trying to compare the
constructs of experienced counselors with counselor's in training. I never
got to the point of creating the grid and having counselors rate role
representatives on the constructs, but I did try this myself with my own
constructs and role representatives..
My current idea is to look at the differences between counselors and
non-counselors. I'm thinking of using video tapes of six individuals and
having subjects first list all the things they notice about the idividual
while they are watching the video, and then, after watching the video, list
all words that they might use to describe the individual. This should elicit
both subbordinate and supeordinate constructs and relate them to each other.
Using video taped people would solve the problem of everyone reacting to
different a different stimulus. I would also have them list the oposite of
each construct on their list when they had finished watching all the videos.
Now the problem is how to make sense of all of the data I will collect.
What I expect to find is that counselors have more subbordinate constructs
and that they have more levels in their construct system. I'm not sure how
this would show itself using a rep grid. It seems to me that the grid
elicits mostly one level of the construct system hiarchy and does not really
give a true pictue of the relationships within the system between the
different levels of the hiarchy. Laddering does provide one method of
looking at the relationships between constructs at different levels, but I
don't think it will work in looking at differences between counselors and
non-counselors unless I would want to start with the big five and ladder down
to test the hypothesis that counselors have more levels in their hiarchy.
I've thought of using cluster analysis or correspondence analysis to
analyse the data.
Well, that's where I'm at right now with my dilemma. I haven't
mentioned my current ideas to my advisor so I don't know what he is going to
say about my newest approach to the dissertation. Sorry for such a long

Caroline Cook
The University of Akron