[R] Testing parallel regression assumption

Medeiros, Rose rosem at ats.ucla.edu
Wed Jun 20 20:37:17 CEST 2007


I would like to test the parallel regression assumption for an ordered
logistic regression model using either a score/LM test, or preferably, a
Wald/Brant test. I have been unable to find an R function (either in the
base or in a package available on cran) that performs either of these
tests. Does one exist that I am overlooking or is there code available
elsewhere to do this?
 
As background, I am using vglm( ) from the library VGAM to fit the
model. I'm estimating the model with code that looks roughly like this: 
vglm(y~x1+x2+x3, data=mydata, family=cumulative(link = "logit", parallel
= T, reverse = T))

Thank you,
Rose
 
 

______________________________________

Rose Anne Medeiros

Statistical Consulting Group

UCLA Academic Technology Services

http://www.ats.ucla.edu/stat/



More information about the R-help mailing list