[R] Coefficients for aliased models
Ross Darnell
r.darnell at uq.edu.au
Mon Apr 24 14:27:50 CEST 2006
I find the reporting of aliased terms in class lm objects as little
inconsistent.
The best way to show this is to give an example.
X <- mvrnorm(10,c(0,0),Sigma=matrix(c(1,0.5,0.5,1),nrow=2))
Xnew <- as.data.frame(cbind(X,X[,2])) # make copy of X[,2]
names(Xnew) <- c("y","X","Xcopy")
model <- lm(y~X+Xcopy,data=Xnew)
coef(model) # reports NA for Xcopy estimate
summary(model) # ditto
coef(summary(model)) # drops NA for Xcopy estimate
Is this intentional? Is there a way of extracting the coefficient table
from the summary output with the aliased term included?
Many thanks
Ross Darnell
More information about the R-help
mailing list