On Tue, Feb 23, 2010 at 2:17 AM, Dieter Menne <dieter.me...@menne-biomed.de> wrote:
> Marianne Promberger-3 wrote: >> >> Yes, I believe you need library(lme4a), the development version of lme4. >> >> .... >> (But then something didn't work with profile() on my particular model >> but I forgot what it was -- haven't had time to pursue this yet) >> >> > > profile() worked for me (="output looked reasonable"), so it looks like it > really is a problem with special model. Well, that and the fact that it is not really well documented. When you have problems with profiling you can try to cut down the step size. If you check the actual code used in Chapter 1 of the book I am writing you will see that I cheat and call profile as pr01 <- profile(fm01, delta = 0.2) That mysterious, and undocumented, delta = 0.2 forces smaller step sizes when constructing the profiles (step size in the sense of the amount of the shift of the parameter being profiled from its current position to its next position). Sometimes large steps cause inability to converge at the new value. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
