[R] (sem assunto)

John Fox jfox at mcmaster.ca
Sun Jan 20 17:01:24 CET 2008


Dear Edison,

There are several approaches you could take. One would be to use a
component+residual plot (see the cr.plots function in the car package, which
has a method for glm objects, including logistic regression), although that
doesn't provide a "test." Another approach would be to fit a nonparametric
regression model, such as a generalized additive model (see the mgcv and gam
packages), with smooth terms in place of linear terms, which can be used to
test linearity.

I hope this helps,
 John

--------------------------------
John Fox, Professor
Department of Sociology
McMaster University
Hamilton, Ontario, Canada L8S 4M4
905-525-9140x23604
http://socserv.mcmaster.ca/jfox


> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of Edison Iglesias de Oliveira Vidal
> Sent: January-19-08 5:31 PM
> To: r-help at r-project.org
> Subject: [R] (sem assunto)
> 
> 
> I have been searching without success to find out how to test for the
> assumption of linear relationship between the indenpendent variables
> and the logit of the dependent variable in multiple logistic regression
> models using R. I shall be glad for any advice.
> 
> best regards,
> 
> Edison Iglesias de Oliveira Vidal,  MD
> Home Care Department, Hospital Israelita Albert Einstein Sco Paulo,
> Brazil
> 
> 
> 
> _________________________________________________________________
> Cansado de espago para ss 50 fotos? Conhega o Spaces, o site de
> relac[[elided trailing spam]]
> 
> 	[[alternative HTML version deleted]]



More information about the R-help mailing list