[Date Index] [Thread Index] [Author Index]

Re: Help me:Statistic-Multivareate Analysis

• To: mathgroup at smc.vnet.net
• Subject: [mg5918] Re: [mg5895] Help me:Statistic-Multivareate Analysis
• From: Steve.Hunka at ualberta.ca (Steve Hunka)
• Date: Tue, 4 Feb 1997 00:03:41 -0500
• Organization: University of Alberta
• Sender: owner-wri-mathgroup at wolfram.com

In article <5d2i9g$cm0 at smc.vnet.net>, cefortel at telepac.pt (Manuel
Rodrigues) wrote:

> Who tell me ??
> 
> I see in a statistical text the following words:
> 
> "VARIMAX rotation".
> 
> Who can help me about that.
> 
> A student of a professional school asked me that.
> 
> Thank you for your help.
> 
> Manuel Rodrigues.
> 
> Cefortel at mail.telepac.pt
> 
>  
Manuel Rodrigues asked what is Varimax rotation. Varimax rotation is an 
orthogonal rotation developed by Henry Kaiser and is used primarily in 
the context of the technique called Factor Analysis. A commonly used 
method in factor analysis resolves a correlation matrix into components
(factors) which are eigenvectors rescaled by the sq-root of their 
eigenvalues. The result gives projections of each variable onto an 
orthogonal axes, and Varimax is a procedure for iteratively maximizing
the variance of these projections. The result gives a new set of projections
which show a clustering of the variables on a single axis (factor) thus
making it more interpretable (Thurstone's Simple Structure notion). The
original reference is a PhD thesis, but the following references may help:
Kaiser, H.F., The Varimax criterion for analytic rotation in Factor 
Analysis, Psychometrika, XXIII (1958), p187-200, and Kaiser, H.F.,
Computer program for Varimax rotation in Factor Analysis, Educ. & Psych.
Meas., 
XIX, 3, 1959. A good text, but out of print, by Stanley Mulaik (Introduction
to Modern Factor Analysis) is also a good source. Steve Hunka, U Alberta,
steve.hunka at ualberta.ca