R help - Jul 2004 - Is there a statistics that can summarize the correlation for more than two random variables?

Hi, R people,

 

I wonder if there is a statistics than can measure the correlation for more
than two random variables, instead of computing the correlation coefficient
matrix. If so, what R package should I use? 

 

Right now I can only think of the mean of all pair-wise correlation
coefficients, e.g., (corr(x,y) + corr(x,z) + corr(y,z)) / 3 for three random
variables (x, y, z). 

 

Thanks a lot,

 

Frank


	[[alternative HTML version deleted]]

On 07/13/04 14:34, F Duan wrote:
>Hi, R people,
>
>
>
>I wonder if there is a statistics than can measure the correlation for more
>than two random variables, instead of computing the correlation coefficient
>matrix. If so, what R package should I use?
One possibility is Cronbach's alpha, which is in the psy
package.  It is describe a little in our "Notes on R for
psychology ..." (linked from the R page below), written befpre
psy was available.

Jon
-- 
Jonathan Baron, Professor of Psychology, University of Pennsylvania
Home page:            http://www.sas.upenn.edu/~baron
R search page:        http://finzi.psych.upenn.edu/

Thank you for your reminding. Could you tell me the addresses of STAT-L and
ALLSTAT lists?

By the way, I found Cronbach's alpha suggested by Prof. Baron might be the
one I am looking for though it's not perfect. 

Frank

-----Original Message-----
From: Peter Flom [mailto:flom at ndri.org] 
Sent: Tuesday, July 13, 2004 16:06
To: r-help at stat.math.ethz.ch; f.duan at yale.edu
Subject: Re: [R] Is there a statistics that can summarize the correlation
formore than two random variables?

This seems more like a STATS question than an R question - asking on a
list like STAT-L or ALLSTAT may result in more replies

Nevertheless, it seems to me that you need to describe (and maybe
decide) what you mean by 'summarize' the correlations.  Certainly the
mean DOES summarize them, but is it the summary you want? Maybe, maybe
not.  Perhaps the median? Or a trimmed mean? Do you want to take the
absolute values of the correlations, or not? 

HTH



Peter L. Flom, PhD
Assistant Director, Statistics and Data Analysis Core
Center for Drug Use and HIV Research
National Development and Research Institutes
71 W. 23rd St
www.peterflom.com
New York, NY 10010
(212) 845-4485 (voice)
(917) 438-0894 (fax)

>>> F Duan <f.duan at yale.edu> 07/13/04 2:34 PM >>>
Hi, R people,

 

I wonder if there is a statistics than can measure the correlation for
more
than two random variables, instead of computing the correlation
coefficient
matrix. If so, what R package should I use? 

 

Right now I can only think of the mean of all pair-wise correlation
coefficients, e.g., (corr(x,y) + corr(x,z) + corr(y,z)) / 3 for three
random
variables (x, y, z). 

 

Thanks a lot,

 

Frank


	[[alternative HTML version deleted]]

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html