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

st: RE: oblong correlation matrices

From   "Nick Cox" <[email protected]>
To   <[email protected]>
Subject   st: RE: oblong correlation matrices
Date   Wed, 17 Nov 2004 10:31:56 -0000

Following Kit's comment, 

I confirm, as the author, that -cpcorr- was written deliberately 
in this way. At the time of writing, I really wanted to put the onus 
on the user to hack -cpcorr- towards what they wanted if they didn't 
like that decision. 

I don't have plans to add this option. A positive reason for 
that is that my -makematrix- already provides the same flexibility. 
The syntax is likely to seem rather complicated, but that's because
the program is much more ambitious. 

Naturally, anyone is free to clone -cpcorr- and modify to taste. 
The Statalist FAQ suggests appropriate etiquette for such 

[email protected] 

Kit Baum
> Wendy wrote
> The cpcorr command seems the simplest method for now.  But it seems to
> use listwise exclusion instead of pairwise for the group of
> correlations, so I am not getting the same results for pairs of
> variables as I get from other weighted correlation commands.  Is it
> possible to use pairwise exclusion with cpcorr -- or is there 
> some other
> reason for getting different results for pairs of variables 
> from pwcorr
> versus cpcorr?
> If you comment out the line of cpcorr.ado
> if `slash' { markout `touse' `rowvars' }
> cpcorr will produce pairwise correlations (and the N reported will be 
> the maximum avaiilable for any correlation). The author of 
> cpcorr might 
> want to add a 'pairwise' option to the routine, � la
> if `slash' & ~`pairwise' { markout `touse'  `rowvars'}
> as I do not see any reason why it would not be sensible to allow 
> pairwise correlations to be generated in this context.

*   For searches and help try:

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