Re: The (Clinical) Point

Lois Shawver (
Fri, 8 Mar 1996 21:58:47 -0800 (PST)


You ask what I think about your theory so far. Let me say that my purpose
here is merely to facilitate you teaching me (and others who listen) about
your theory. If I say things that sound critical, it is because I am
trying to imagine how you are thinking and I am presuming that my sense
that you are saying something foolish is a mistake that you can correct.
I have disdain for premature criticism of anyone's theory. It's an
unfortunate western tradition. I hope my series of questions are working
to help you feel you are getting your ideas out. If not, please
collaborate with me in shaping the discourse. It is just that it is
difficult for you, or for me, to elaborate a theory without someone to ask
questions or make comments.

Let me say, though, that it has been my experience that sometimes when
people work through their ideas with someone who questions them closely,
they sometimes see things that they want to change. If so, then I would
chalk that up to something positive and I hope you would, too.

One more thing. I am going to work through your material below, and to
make sure your remarks are distinguished from mine, I'll insert >
in the margin by your notes.

You say:

> The Core is the essential phenomenon that
> the method of corresponding regressions
> measures. The method of corresponding
> regressions should be useful in helping us
> determine which constructs are made up of
> other constructs.

> For example, lets say you have a client who
> cuts her arms. You want to study the series of
> constructs that add up in her mind to "cut my
> arm." You collect a group of experiences in the
> patients life- such as "talking with boy friend
> after school", "hearing mom and dad fight", etc.
> Get about 50 of these. Now ask the client to rate
> these experiences along the constructs that you
> think may be clinically significant. Lets say they
> are A: "makes me feel angry", B:"makes me feel
> empty"and C: "makes me want to cut myself".
> There are no doubt many more that a clinician
> could mention and you could use as many as
> seem of clinical interest.

So, I have a list of experiences that the client has told me about, and I
choose the constructs that the patient uses for the rating. Let's say
the experiences are
1) a talk with her boyfriend after school
2) hearing mom and dad fight
3) having an argument with her sister

Can we refer to these as experiences 1, 2 and 3? And shall I presume
that you mean a concrete individual experience? Not a "type" of experience?
I mean, we are going to ask her to rate these constructs for a particular
talk she has with her boyfriend after school, last saturday say?
Or do you think this makes a difference? Could it be "all talks she has
with her boyfriend after school" regardless of content or mood, etc.?

Now I choose the constructs and have her rate them. I set up a grid
based on
A: "makes me feel angry",
B:"makes me feel empty"
C: "makes me want to cut myself".

Together with the experiences it looks like this


She rates experience 1 (talk with boyfriend after school) according to
whether it makes her feel angry (A), and so forth. Is this right?
Do I use a 9 point scale for this? Or what? I end up with something
like this?

1. 2 4 9
2. 5 5 8
3. 8 3 4

With these scores representing ratings?

You say:
> Have the client rate the 50 experiences on the
> constructs. Do the corresponding regressions
> analysis. It should tell you which of the
> experiences add up to "Cut myself", if any.

In regression we would try to discover a formula to predict this
dependent variable ("cut myself") from the weighting of the ratings (I get
angry when I talk to my boyfriend after school). Right? If the
regression studies come out well, then you would be able to say things
like "If she gets really, really angry when she talks with her boyfriend
after school, but does not feel "empty" on these occasions ( and so forth
for all the variables), then we know she is going to cut herself (be a
person who cuts herself generally? Or cuts herself on a particular
occasion?) I see a lot of fuzziness that I need clearing up here to
really follow you. I want to know, most of all, if you are talking about
events that are rated, or something more general, and if you are
predicting events fo cutting herself, or something more general.

You add:

> I do not know the psychology of "cutters".
> Clinicians may know what goes through their
> minds. But given enough clients, with enough
> problems, sooner or later all run into a case
> where the client's constructions are obscure and
> perhaps unknown to anyone. You could spend
> many hours or even years trying to figure out how
> it all goes together. In theory, at least, a grid and
> corresponding regressions should make the
> information available relatively immediately, saving
> you time and the danger of making a mistake in
> figuring out the client's constructions.

Now it occurs to me that maybe you are not thinking of doing these
regressions for particular people, but for samples of people. Is that
right? Maybe in this case it would be for samples of people who cut
their arms? And you would want to find out which kind of pattern of
construct ratings of events would predict arm cutting? Then you would
give this information to a clinician who had a arm-cutting patient and it
would tell the clinician that the patient is "likely" to, for example,
get very angry at boyfriends when talking after school but not feel empty
in these situations?

> It could require more than 20 minutes of your time
> to gain these skills, but the benefit to you and
> your patients, in theory, should far exceed your
> investment in understanding the psychometrics.
> And a computer program could be prepared by
> someone like me to do the actual math for you.

Good. I alwys like the computers to do the math. BUT there is the
danger, of course, that someone will use them to do the math without
understanding what the math means. It's the downside of computerized
psychometrics -- but then there is the wonderful upside. Right? Who
wants to do these things by hand themselves?

> Let your imagination run. Would it be useful to
> very quickly and explicitly see what makes
> your client's "tic". You do not always have much
> time before your client loses faith in you as a
> therapist. They sometimes think good doctors
> can just about read their minds. Take too long,
> and they may dismiss you and go back to more
> reliable ways of construing, like cutting their
> arms.

Well, it's nice to have some help in undestanding patients. Is the idea
that I would then use the information to offer a diagnostic statement to
the patient in the therapy? Saying something like, "You are a person who
cuts yourself when you feel angry with your boyfriend after school?" Can
you tell me more what you have in mind here?

> Or they may linger, waiting for you to
> figure out what leads up to the cutting. Sooner or
> later, she might get unlucky and hit an artery. And
> then she is dead. Or he jumps from a building or
> she divorces her husband or he hits his kid or
> what ever.

Better to understand them correctly? So that you can explain it to them
in the hopes that, feeling understood, they will desist?

Interesting ideas.

..Lois Shawver