Anthropac is a software program developed by Steve Borgatti. His
phone number is (803)777-3123. He provides excellent technical
support, and would be happy to answer anyone's questions, I'm sure.
I think that the Triads questionnaire would be an ideal tool for
examining cultural differences in the way that emotions are perceived, or
as Jay put it, how "close together" certain emotions are. In fact, that's
exactly the information that the Triads questionnaire would give you.
This ties in rather nicely with the discussion of culture and how it can
best be defined (which came out of that "coercion" debate).
If culture is defined as a meaning system, a way of viewing things that
is similar to that of others in your "culture" (and by culture, I don't just
mean ethnic group--I mean groups of people who agree on the meanings of
a certain domain, such as emotions, that you're interested in.).
So, a triads questionnaire would present items (words, emotions, whatever
you're interested in) in groups of three, and each respondent will be
asked to circle the item that he/she thinks is most different in meaning
from the other two. For instance, if you give them the triad:
HOUSE CAT DOG
Most people would choose HOUSE as being the most different in meaning.
(However, when you're using things such as emotions, the task will
be much more challenging!)
The first thing you have to do, when analyzing Triads data, is to
run a consensus ayalysis. This tells you whether the respondents are
all of the same culture--which means that they have agreed upon the
"correct" answers to the questionnaire. You can actually map out,
using MultiDimensional Scaling, how closely you respondent's answers
correspond to each other. If they form a nice, tight cluster, then
you can assume they are all in the same culture--I believe Steve Borgatti
refers to this as "drawing cultural boundaries."
Once your cultural domains have been established using Multi Dimensional
Scaling, then you may go on to look at what meaning clusters your
respondents have agreed upon. Again, you're using MultiDimensional Scaling
and (in addition) Cluster Analysis. What this gives you is a map
of what words are considered to be similar in meaning. So, those words
or items that are perceived by the respondents as being very similar will
appear very close to each other on the map, and those that are considered
very different in meaning will be further away. What you will (ideally)
come out with are "clusters" of items that are considered similar.
So, if you're studying emotions, maybe happiness, joy, and love will
all be clustered together (who knows?).
This is a very different technique that is beginning to be used more
frequently by medical anthropologists and cultural psychologists.
Since it is so different from what is conventionally used in Psych,
it is sometimes difficult to explain--so I hope I've done an adequate
job here. If anyone has questions, feel free to ask.
You know, I think we have a very interesting research project here!
I once did research (based somewhat on Eckman's research, which I think
was discussed earlier by this group) on gender differences in decoding
affect from facial expressions. I think, however, that this is a much
more interesting study! Whereas my study used the "universal" emotions
identified by Paul Eckman, this gets at something much more meaningful!
After all, our perception of something as subjective as emotions
is bound to be cultually based! And you could look at gender differences--
differences between any two groups, really.
Well, food for thought!
-Pam
Pam Schulze
PAS94003 who-is-at UCONNVM
UNIVERSITY OF CONNECTICUT