Vikki,
The formula you used for std. error of C is not correct. C is not a simple
per-observation proportion.
SD in the output is the standard error of Dxy. Dxy = 2(C - .5). Backsolve
for std err of C.
Variation in Dxy or C comes from the usual source: sampling variability.
You can also see this by sampling from the original dataset (a la
bootstrap).
Frank
vikkiyft wrote:
>
> Dear R-help,
>
> This is an example in the {Hmisc} help manual in the section of
> rcorr.cens:
>
>> set.seed(1)
>> x <- round(rnorm(200))
>> y <- rnorm(200)
>> round(rcorr.cens(x, y, outx=F),4)
> C Index Dxy S.D. n missing
> uncensored Relevant Pairs Concordant Uncertain
> 0.4831 -0.0338 0.0462 200.0000 0.0000
>
> 200.0000 39800.0000 19228.0000 0.0000
>
> That S.D. confuses me!!
> It is obviously not the standard deviation of x or y.. but there is only
> one realization of the c-index or Dxy for this sample dataset, where does
> the variation come from..?? if I use the conventional formula for
> calculating the standard deviation of proportions: sqrt((C Index)*(1-C
> Index)/n), I get 0.0353 instead of 0.0462..
>
> Any advice is appreciated.
>
>
> Vikki
>
-----
Frank Harrell
Department of Biostatistics, Vanderbilt University
--
View this message in context:
http://r.789695.n4.nabble.com/what-does-the-S-D-returned-by-Hmisc-rcorr-cens-measure-tp3329609p3329899.html
Sent from the R help mailing list archive at Nabble.com.
______________________________________________
R-help@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.
