Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: RE: Factor analysis and dummy variables.


From   "Nick Cox" <[email protected]>
To   <[email protected]>
Subject   RE: st: RE: Factor analysis and dummy variables.
Date   Fri, 10 Dec 2004 18:18:36 -0000

First, you said "factor analysis" in your 
title, but "PCA" in your text. I will stick 
with PCA. 

Second, one view -- there are others -- of PCA 
is that it is just based on taking eigenvalues
and eigenvectors of (positive semi-definite 
(symmetric)) matrices. So, it is like any other 
calculation: it could be the right thing to 
do, depending on the circumstances and what 
kind of answer you want. Otherwise put,
PCA is a transformation technique, not a modelling
technique. Just like taking logs, it could be
helpful, or it could tell you little. That 
is not quite right or wrong. 

To repeat, there are are other views. 

I've heard of applications of eigen* 
technique to 0-1 matrices, and no doubt 
there is a literature. There's probably 
some corner of physics in which this is 
a bread-and-butter technique. 




© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index