CSU (csu@brain.wph.uq.oz.au)
17 Oct 1995 08:36:51 -0500


It sounds as if you now have lots of ideas and Travis has
suggested the use of multidimensional scaling (which I have found
interesting). My suggestion would be to go back to your initial
question, which I think concerned youth and sexual risk taking.
My view is that the feasibility study should not be overly
complex but rather be designed to provide access to how youth
construe risk so as to guide you in examining your question.

I would probably supply some situations which I think are high
and low risk and let participants suggest a few of their own.
These could be rated, ranked or as Travis suggested pairs of
situations could be compared in terms of (dis)similarity.
Regarding this latter approach see: Schiffman S, Reynolds M,
Young F (1981). Introduction to Multidimensional Scaling:
Theory, Methodology and Applications. Academic Press, Orlando.

Laddering would be a useful means to examine more fundamental
constructs, whereas the other approaches you mention might be
more fruitfully used after the feasibility study. My reason for
saying this is that any form of grid can easily take an hour.
When you add the other approaches you mentioned you need to
consider the logistics and stamina of all the parties involved.

Regarding the use of subgroups, I have found this seemingly
simple matter quite complex, for example; what dimension do you
think these groups differ on (secondary schools, first year
university and juvenile offender populations)? Without a large
sample you may find drawing conclusions about group differences

Regarding the use of the 2 scales you mentioned I have no
knowledge of these. If you wished to correlate these scale
scores with grid data then a standard grid would seem most useful
(such a grid would be somewhat akin to a questionnaire)
supplemented by a narrative approach. There are many
possibilities, my basic advice is however to keep focused on the
question of interest and keep it achievable,


Bob Green