glm throws erroneous(?)'names attribute must be the same length as the vector' (PR#144)

[wagman@enteract.com]

Full_Name: Barnet Wagman
Version: 63.2
OS: Linux
Submission from: (NULL) (207.229.172.168)


After a number of interations, glm(family=binomial) throws an error
	'Error: names attribute must be the same length as the vector'
The same model works in lm() with the same data frame, so presumably it is not
misspecified.

In the iterations prior to the error, glm() generates the warnings
	'Warning: fitted probabilities of 0 or 1 occurred'

The fit is converging prior to the error.

Could the problem involve a singularity and the wrong error message?

FYI The model contains 34 variables in around 11,000 cases.
The command used is
'glm(formula=...,data=...,family=binomial,control=control.glm(epsilon=0.0001.maxit=50,
 trace=TRUE))'
Changing epsilon and/or explicitly setting 'start' does not effect the problem.


Thanks for your help.

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._