Skip to content

Factor Analysis in MASS4

4 messages · Ko-Kang Kevin Wang, Brian Ripley, John Fox

#
Hi,

I had a look at the MASS4 scripts in the MASS package, in Ch 11.3 Factor
Analysis, there is a section of codes like:

  data(ability.cov)

  ability.FA <- factanal(covmat = ability.cov, factors = 1)
  ability.FA
  (ability.FA <- update(ability.FA, factors = 2))
  #summary(ability.FA)
  round(loadings(ability.FA) %*% t(loadings(ability.FA)) +
             diag(ability.FA$uniq), 3)

Unfortunately I still haven't received the book I ordered, so I can't look
this up.

Two questions:

1) What does the update() do?  I mean, what happens if I replace it with 
     factanal(covmat = ability.cov, factors = 2)

2) What does the last command, the formulae in round() mean?  I tried it
and it produced a matrix that looks kind of like correlation matrix of
some sort... 


Cheers,

Kevin

------------------------------------------------------------------------------
Ko-Kang Kevin Wang
Postgraduate PGDipSci Student
Department of Statistics
University of Auckland
New Zealand
Homepage: http://www.stat.auckland.ac.nz/~kwan022


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
#
On Thu, 29 Aug 2002, Ko-Kang Kevin Wang wrote:

            
That's what it does.  update always recalls the original call with the
changes as given.
It is the fitted correlations.
#
I see.

Just out of interest, is it possible to do a regresion analysis on the
factors obtained from the factor analysis?
On Thu, 29 Aug 2002 ripley at stats.ox.ac.uk wrote:

            
Cheers,

Kevin

------------------------------------------------------------------------------
Ko-Kang Kevin Wang
Postgraduate PGDipSci Student
Department of Statistics
University of Auckland
New Zealand
Homepage: http://www.stat.auckland.ac.nz/~kwan022


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
#
Dear Kevin,
At 10:03 PM 8/30/2002 +1200, Ko-Kang Kevin Wang wrote:
Dear Kevin,

Factor scores are variables, so, as a mechanical matter, you can use them 
in a subsequent analysis. (You won't get factor scores if you start with a 
covariance matrix, as opposed to a data matrix.) Factor scores can have 
large measurement-error components, however, which causes problems if you 
use them as explanatory variables in a regression.

If your goal is to do a regression using the factors, and you can specify 
in advance which variables load on which factors, you might consider 
estimating the factor loadings and regression coefficients simultaneously, 
e.g., with the sem (structural-equation model) function in the sem package.

John
-----------------------------------------------------
John Fox
Department of Sociology
McMaster University
Hamilton, Ontario, Canada L8S 4M4
email: jfox at mcmaster.ca
phone: 905-525-9140x23604
web: www.socsci.mcmaster.ca/jfox
-----------------------------------------------------

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._