Skip to content
Prev 22320 / 398502 Next

More on problem with polr

(those concerned about finding out who I am will find clues carefully
hidden towards the end of this message)



Briefly, I'm using polr (from the MASS library) to re-run an analysis I
did about 18 months ago, with a previous version of R (I'm now using
1.5.0, I think I was using 1.2.x or 1.1.x before).  It ran fine before,
but now I'm getting an error.  David Duffy suggested I post a minimum
data set to demonstrate the problem, so here is one which shows the
error - not having the old versoin of R, I can't check whether it would
work with that:
Error in optim(start, fmin, gmin, method = "BFGS", hessian = Hess) : 
        non-finite value supplied by optim
The full data set is much bigger (over 380 observations), but I can
provide it if someone wants to look at it in more detail.  The full data
set does have "holes" in it (i.e. combinations of levels with no
observations), but similar data sets with the same sorts of holes run
fine.

Thanks in advance for any help!

Bob