One wouldn't. In fact, I rather suspect that factor analysis provides
a way of mathematizing social bias rather than getting rid of it,
given its usual post-hoc nature.
> I don't get it.
Join the club!
> It is easy to use the sausage metaphor to talk about factor analysis.
> [It's hard for me to refrain from using the word BALONEY rather than the word
> SAUSAGE!!!]
> One puts ground meat into the computer, and it comes out in links!
> Nice and neat, and useful for generating speculation. But it is still packaged
> ground meat!!! Sausage -- ughhh, I needed to bite my tongue once again I
> almost started typing that B word.
Great metaphor! And one that leads me to an area I've been reading up
on: Facet Theory, devised by the great Louis Guttman. Anybody else
out there looking at this approach? Essentially, it not only requires
us to define our conceptual space a priori, but it is more general
from an analytic perspective in that the restrictive and oft-violated
assumptions of FA can be avoided. Likewise, the approach spurns what
I call the "variable soup" approach to devising measures where tons of
items are sifted, and - no matter how conceptually (un)related they
are - get lumped into clumps on the basis of statistical properties
that we can't necessarily separate from chance factors or subjective
bias. Rather, it uses simple set theory to derive items that make
conceptual sense and usually are pretty good about loading into
predefined spaces.
Facet theory forces us to make explicit our constructs ahead of time,
and obliges us to examine the extent to which the data conform to our
conception of the shape it should take. In other words, we prep the
packages for the sausage links beforehand, and hope that our ground
meat fits into them!
{Sam Shye recently put out a book with Ingwer Borg on the topic, "Facet
Theory: Form and Content," for interested readers. It's available
from order@sagepub.com for about $38 US.}
cheers,
((((((((((((((((((((((((((((((((((((((((((((((((((((((((((
Travis Gee () tgee@superior.carleton.ca ()
() () ()()()()
() () ()
() ()()()()()()()()()
"In science, the more we know the more extensive the
contact with nescience." -Spencer
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%