[R] Gamma for ordinal trends

Peter Dalgaard p.dalgaard at biostat.ku.dk
Fri Sep 2 13:37:39 CEST 2005


Wuming Gong <wuming.gong at gmail.com> writes:

> Is it possible to use delta method to evaluate the standard error of
> Goodman-Kruskal gamma and then Wald test to evaluate the significance
> of association?
> 
> Wuming

Probably better to actually read the paper(s) by Goodman&Kruskal
(JASA(1972), vol.67, 415--421, &c). AFAICS, that basically *is* the
delta method, but you'd likely want a score test, not a Wald test
(i.e. calculate the variance under the null).
 
> 
> On 31 Aug 2005 13:42:27 +0200, Peter Dalgaard <p.dalgaard at biostat.ku.dk> wrote:
> > Jonathan Baron <baron at psych.upenn.edu> writes:
> > 
> > > On 08/31/05 17:46, Wuming Gong wrote:
> > > > Dear list,
> > > >
> > > > Are there any functions for calculating gamma (and its standard
> > > > error), which measures the association of ordinal factors in I x J
> > > > contingency table. I did a RSiteSearch but did not find any clues...
> > >
> > > You have to look for Goodman-Kruskal gamma.  It is a bit
> > > obscure.  It is rcorr.cens in the Hmisc package.
> > >
> > > The significance test is the same as for Kendall's tau, according
> > > to some books.
> > 
> > Well, it would be if we handled ties in Kendall's tau correctly...
> > 
> > > I don't know about standard error.
> > 
> > --
> >    O__  ---- Peter Dalgaard             Øster 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
> >
> 

-- 
   O__  ---- Peter Dalgaard             Øster 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




More information about the R-help mailing list