R help - Oct 2005 - cor() function, method="spearman"

Hello,

Does anyone know if the cor function, when method = "spearman",
returns a correlation coefficient corrected for any ties in the ranks of the
data?  I have data with quite a few ties and am thinking that I should use a
calculation of the coefficient corrected for ties, but before I try and code
this calculation myself, I thought I should check whether or not cor()
automatically does this.

Thank you!
Joanna McGarvie

jlmcgarv at sfu.ca writes:
> Hello,
> 
> Does anyone know if the cor function, when method = "spearman",
returns a correlation coefficient corrected for any ties in the ranks of the
data?  I have data with quite a few ties and am thinking that I should use a
calculation of the coefficient corrected for ties, but before I try and code
this calculation myself, I thought I should check whether or not cor()
automatically does this.
> 
> Thank you!
> Joanna McGarvie
The tie-corrected spearman correlation is the correlation of the
tie-corrected ranks, so yes, it does do that. 

Previous versions of R had an issue with p-values in cor.test, where
it used the exact distribution of the test for non-tied data, even
when ties were present. This was straightened out in 2.2.0, where it
now uses an asymptotic formula instead.

-- 
   O__  ---- Peter Dalgaard             ??ster Farimagsgade 5, Entr.B
  c/ /'_ --- Dept. of Biostatistics     PO Box 2099, 1014 Cph. K
 (*) \(*) -- University of Copenhagen   Denmark          Ph:  (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)                  FAX: (+45) 35327907