Mob Rule
Fri, 15 Mar 1996 09:43:18 -0500
Content-ID: <>
Content-type: text/plain

Attach file.
Content-ID: <>
Content-type: text/plain;
Content-Transfer-Encoding: quoted-printable

Mob rule in factor analysis can come
about when a particular factor is over
sampled in construct elicitation. The
assumption that the most important
constructs are the ones that are replicated
in elicitation is risky. A person could give
numerous variations on some theme that
may or may not have much weight in his
construct system. Or, a group could waste
a lot of resources tauting variations of a
method that is merely socially validated,
thinking that the mob has factored out the
dissenting variable. =

Kelly was aware of this problem and
recognized the need to factor by a =

priori dimensions. Read Kelly's PPC
p.295. Kelly's method for confirmatory
factoring was conceptually akin to the =

more general and elegant method of =

multiple group factor analysis (MGFA). =

In MGFA the experimenter designates
the factor, rather than letting mob rule
make the determination. The designation
procedes on theoretical grounds. When
dealing with a psychopath, the psychologist
may supply the construct "makes me feel
guilty." The doctor can then see how much
of the variance is accounted for by the guilt
factor, even if the psychopath does not have
much to say about it in the construct
elicitation. =

Multiple group factor analysis is an
option in Circumgrids. The user can
designate a factor by defining a =

particular construct as the factor or
by composing a factor from combinations
of constructs. More than one factor
can be composed. Circumgrids tells if =

the factors correlate and it will provide both
oblique and orthogonal loadings. Circumgrids
also allows the user to factor first with
multiple group and then to use principal
components on the residual of the initial

One wonders if the person who reviewed
Circumgrids for Bob Neimeyer's journal
even booted the program up. As I recall
it was dismissed as being of not much
use because it did not store the grids that
were input on to disk. Perhaps they
should have read their Kelly closer and
been more concerned with what Circumgrids
can do rather than what it does not do. It is
typical of the Kelly crowd to dismiss well
conceived and potentially very fruitful
endeavors by referencing some incidental
shortcoming. Mean while, the little people
pay, in one way or another, for the leaders' =

disregard for the whole truth.
So, what do you think about mob rule
and multiple group factor analysis of
grids? Circumgrids has been out for
over ten years and I received over 500
requests for it. Why hasn't it been the
subject of any of the the mob of journal
articles and book chapters?