Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


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

st: RE: RE: Spearman correlation with adjustment


From   Nick Cox <n.j.cox@durham.ac.uk>
To   "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu>
Subject   st: RE: RE: Spearman correlation with adjustment
Date   Thu, 6 Oct 2011 17:54:08 +0100

To be more positive and more specific: 

Much fancier models are popular, but I would start by looking at (difference between methods) and look for relationships with weight and age. 

That in itself may show that (difference between methods) is not the best metric. 

Grouping ages just throws away information. 

Normality, even conditional normality of the response, is nowhere essential. 

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


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Nick Cox
Sent: 06 October 2011 11:58
To: 'statalist@hsphsun2.harvard.edu'
Subject: st: RE: Spearman correlation with adjustment

My prejudice is that this isn't a very fruitful direction to be looking.  

Either you think in terms of modelling this properly, or you think in terms of different correlations. If they are very similar, no problem arises; if they are very different, it is hard to see that the idea of a single underlying correlation makes much sense. 

All that said, if the underlying issue is do methods agree, then arguably _concordance correlation_ is what you need: -findit concord-. 

A knock-down example is that corr(y, by) = 1 for _any_ positive b. In words, correlation measures linearity, not agreement. 

I wouldn't be that put off correlation by non-normality. You could always bootstrap it. 

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


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of cecilia sam
Sent: 06 October 2011 11:48
To: statalist@hsphsun2.harvard.edu
Subject: st: Spearman correlation with adjustment

Hi all,

I need to validate a method against another method. Since the data are
not normally distributed, I use spearman correlation using syntax
"spearman" to find correlation, and bland and altman plot to generate
the limit of agreement. I know that there are some confoundings, such
as weight and age. To adjust age, I have classified them into groups
and analysed the correlation for each group, while I want to present
an overall correlation. My question is:

Is there any syntax or method that I can use to adjust a single
continous (Not categorical) variable, or even adjust all of them at
once?

Many thanks.
Cheers from Cecilia
*
*   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/

*
*   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/

*
*   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   |   Site index