Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: Factor Analysis: which explained variance?


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: Factor Analysis: which explained variance?
Date   Mon, 21 Dec 2009 14:40:40 -0000

Why this stance? 

You could justify your choice on scientific or substantive grounds, if
you have any. 

You could try other choices and show that 1 produces the best results --
if that's true -- or alternatively that the choice is immaterial -- if
that's true. Alternatively, trying other choices might indicate that the
choice of 1 is not a good idea. 

P.S. -polychoric- is a command, not an option. It is also a user-written
command, so you should say where it comes from. 

P.P.S. the whole notion of variance is perhaps a little suspect when the
originals are indicator variables. 

Nick 
n.j.cox@durham.ac.uk 

Francesco Burchi

I have a question on factor analysis, using the -polychoric- option
because I have 4 dummy variables:
I run this program:

polychoric Var1 Var2 Var3 Var4
matrix R = r(R)
factormat R, n(6926) ipf   factor(1)          
rotate, horst blanks(.3) 
predict FACTORROTATE

This way, I created an index based on the first factor. I needed this
variable to be included as coviariate in my regression. My question is:
how
do I defend the fact that I have used one single factor? According to
the
initial choice of number of factors I get a different value in the
"proportion" of the variance accounted for by the factor. 
I add that I am using Stata 10.


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index