R help - Jun 2011 - Somers Dyx

Hello R Community,
 
I'm continuing to work through logistic regression (thanks for all the help
on score test) and have come up against a new opposition.
 
I'm trying to compute Somers Dyx as some suggest this is the preferred
method to Somers Dxy (Demaris, 1992).  I have searchered the [R] archieves to no
avail for a function or code to compute Dyx (not Dxy).  The overview of Hmisc
has mention of Dyx for the rcorr.cens function but this appears to be a misprint
because the manual states the function finds Dxy.  Peng and So (1998) state that
the Dyx is easily calculated in SAS (which tells me the same is possible for
[R]).
 
Yang, K., Miller, G. J., & Miller, G. state that:
 
(Tau-b)^2=Somers Dxy * Somers Dyx 
 
…so maybe an approach would be to write a function that is:
 
Somers Dyx<-(Tau-b)^2/Somers Dxy  
 
I just don't want to waste time if this is incorrect logic and/or
there's an easier way to calculate this thing; perhaps there’s a ‘golden’
function already created in an [R] package that I'm overlooking.
 
Thanks in advance,
Tyler   		 	   		  
	[[alternative HTML version deleted]]

Tyler,

The rcorr.cens, rcorrp.cens, and somers2 functions compute Dxy.  That is a
typo in the overview, which we'll fix.  Dxy is the appropriate measure for
binary Y, I think.  It is a simple translation of the ROC area and does not
penalize for ties on Y.
Frank

Tyler Rinker wrote:
> 
> Hello R Community,
>  
> I'm continuing to work through logistic regression (thanks for all the
> help on score test) and have come up against a new opposition.
>  
> I'm trying to compute Somers Dyx as some suggest this is the preferred
> method to Somers Dxy (Demaris, 1992).  I have searchered the [R] archieves
> to no avail for a function or code to compute Dyx (not Dxy).  The overview
> of Hmisc has mention of Dyx for the rcorr.cens function but this appears
> to be a misprint because the manual states the function finds Dxy.  Peng
> and So (1998) state that the Dyx is easily calculated in SAS (which tells
> me the same is possible for [R]).  
>  
> Yang, K., Miller, G. J., & Miller, G. state that:
>  
> (Tau-b)^2=Somers Dxy * Somers Dyx 
>  
> 
so maybe an approach would be to write a function that
is:
>  
> Somers Dyx<-(Tau-b)^2/Somers Dxy  
>  
> I just don't want to waste time if this is incorrect logic and/or
there's
> an easier way to calculate this thing; perhaps there?s a ?golden? function
> already created in an [R] package that I'm overlooking.
>  
> Thanks in advance,
> Tyler   		 	   		  
> 	[[alternative HTML version deleted]]
> 
> 
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 

-----
Frank Harrell
Department of Biostatistics, Vanderbilt University
--
View this message in context:
http://r.789695.n4.nabble.com/Somers-Dyx-tp3593119p3593649.html
Sent from the R help mailing list archive at Nabble.com.