> Adrian Teo wrote:
>
> Vince is essentially correct in his claim that factor analysis (FA) does
> not interpret. To be more precise, FA is not even a single method, but it
> refers to a collection of related algebraic manipulations which is part of
> a large family of analyses of covariance matrices. FA can be exploratory,
> where you allow the data to "speak for itself" or confirmatory, where you
> test particular hypotheses about the underlying structure of the data set.
> The data reduction approach that Vince and I think Hammond is talking
> about sounds like Principle Components Analysis, an exploratory approach.
> For a quick, easy and reliable reference, Sage University has a series of
> booklets on statistical procedures and there is one by Kim and Mueller
> that introduces this class of analysis. This discussion has sometimes
> given the impression that FA is some complex, exotic statistical approach
> that few understand, when in fact, it is commonly used and discussed in
> personality assessments, aptitude and achievement test constructions, and
> diagnostic measures. Many graduate students in various branches of
> psychology take such a course in their second or third year.
>
[Hammond]
This is all true and of course might have been copied off the
first page of any (modern) textbook on Factor Analysis.
However, you don't have to know anything about Factor Analysis
mathematics to figure out what the scientific proof of God is...
all the F.A. work has already been done and it took thousands of
scientists a hundred years to do it.
The bottom line is that all evidence converges to E,N,P,g
at the 2nd order, and that these 3 dimensions (eigenvecors, Factors)
are caused by the gross macroscopic structural geometry of the
brain (Hammond 1994). Most of you have heard of "Sperrian Lateralization",
well, that's just the 1st-axis, turns out there are 2-more just like
Sperry's axis (Bell-Magendie, and the Neuraxis itself). This
causes E,N,P. When you add IQ to that, which is a "time dimension"
(mental speed = IQ), then you have 3-space axes and 1-time axis,
and SURE ENOUGH, you can show how they are physically, mechanically,
causally caused by the 4-axis of space-time (X,Y,Z,t so called)
of real space. Now let me repeat that, X,Y,Z,t PHYSICALLY
MECHANICALLY CAUSES E,N,P,g in Psychology.... they are not just
"similar", there is a direct chain of physical causation (brain
geometry is caused by space geometry).
OK, from there (all of which has now been overwhelmingly proven),
it is only a trivial step to the scientific proof of God.... in fact,
all you do is factor the 4x4 correlation matrix of E,N,P,g (which
can only have a single factor), and that factor is GOD.
QED, God exists.
> And BTW, the ENP by Hans Eysenck is only one of several models that
> reduces personality measurements to common factors. A much more widely
> accepted model is the Big Five (as the name suggests, there are not 3, but
> 5 factors). Eysenck's ENP has not been consistently supported in the
> literature.
[Hammond]
A little bit of knowledge is dangerous (fortunately
not dangerous enough). Turns out 3 of the Big-5 dimension
ARE IN FACT identical to Eysenck's E,N,P.. and the other
two are simply two diagonals in the E-N plane. I have published
the proof of this in the peer reviewed literature (Hammond 1994):
HAMMOND G.E. (1994) The Cartesian Theory: Unification of
Eysenck and Gray, in: New Ideas In Psychology,
Vol 12(2) pp 153-167, Pergamon Press
And it reconciles ALL of the known and published F.A. models in
the literature including Eysencks Giant 3, AVA 4, Big-5, Brand's
Big-6, K&J's 7F, Saucier's 9F, and finally Cattell's 13F 2nd
order model. as is proven by Hammond (1994), ALL OF THESE MODELS
are just the various symmetric redactions of the 13-Symmetry axes
of the common cube. This is proven to two decimal point accuracy
by simply taking the arcosine the published correlation coefficients
and showing that form said geometrical structure. Cattell, the old
master, is of course the only one to have actually resolved all
13 actors of the cube. The Big-5 was discovered by Norman, Goldberg,
Costa & McCrae etc. who are basically academic types equipped
with a desktop computer, commercial Factor Programs, and readily
available captive test subjects (students, patients) etc.
Of course the stronger the redaction (lower the number of redacted
axes) that you take in the cube, the MORE ROBUST the result, since
you're forcing all of the variance into fewer factors. In fact,
Eysenck's-3 (ENP) is the STRONGEST simply because of this, while
Cattell's "all 13 cubic axes" is the hardest to clearly resolve
because the variance is spread among all 13 axis.
BTW, you can look at a cube and count the axes; 3-Normals, 4-Body
Diagonals, and 6 "face diagonals" (see any geometry book). So,
3+4+6=13.
For your reading convenience and enjoyment I have placed a
fully illustrated facsimile copy on my website at:
http://people.ne.mediaone.net/ghammond/cart.html
You can read it at your leisure.
>
> Hope this helps.
>
> Adrian.
>
Thanks, George
-- BE SURE TO VISIT MY WEBSITE, BELOW: ----------------------------------------------------------- George Hammond, M.S. Physics Email: ghammond@mediaone.net Website: http://people.ne.mediaone.net/ghammond/index.html -----------------------------------------------------------
This archive was generated by hypermail 2b29 : Sun May 20 2001 - 18:29:47 EDT