Re: Applications in Social Sciences?

Burton L. Alperson (balpers@calstatela.edu)
Fri, 28 Nov 1997 19:49:48 +0100 (MET)

Jonathan Larson <eltechno@clear.lakes.com> wrote in article
<199711110826.CAA09461@Clear.Lakes.com>...
> In article <199711051659.RAA33368@titan.rz.Uni-Osnabrueck.DE>,
> fuzzy-mail@dbai.tuwien.ac.at wrote:
>
> > Fuzzy concepts seem to be adequat to model social behavior.

<snip>

> > Holger Bruch
>
> Mr. Bruch
>
> I too am trying to use fuzzy in social science applications

<snip>

> http://clear.lakes.com/~eltechno/FUZZY.html/
>
>
Several years ago, I developed a Boolean calculus to operationalize
concepts such as agreement, understanding, expected agreement, validation
of self concept, expected validation of self concept, etc., and the
accuracy of several of the above determinations. (The calculus is also
relevant to group phenomena such as stereotyping, ethnic identification,
acculturation, etc.) I have recently extended the calculus by converting
all determinations to fuzzy logic determinations (I haven't published on
the conversion yet).

I am slowly converting several old data sets to see if the fuzzy extension
improves the interpretation. To some extent, you are always pulling
yourself up by your own bootstraps when you try to compare the utility of
two different metrics, but so far the results are encouraging. I am finding
that when the results are nonsignificant, there is no clear advantage of
one method over the other. However, when the results are significant, so
far, the fuzzy determinations always account for more variance than the
older Boolean equivalents. The fuzzy determinations have the additional
advantage of allowing metric analysis of individual items, a luxury not
afforded by the Boolean techniques. They also allow you to deal with
responses from subjects without dichotomizing them.

If anyone is interested in corresponding on issues such as these (or any
other social science applications of fuzzy logic) I would welcome the
opportunity.

Burt