[R-sig-ME] error from clmm

Rune Haubo rhbc at imm.dtu.dk
Tue May 22 15:51:01 CEST 2012


Malcolm,

It happens that I have been doing some work tracking down that
particular error message in the last couple of days, and I believe I
have found a solution, at least for the most common link functions. I
have committed the changes to R-Forge, so you should be able to
install with install.packages("ordinal",
repos="http://R-Forge.R-project.org") when the package has been build
- make sure you get at least version 2012-05-22 corresponding to
revision 78. I will submit an updated version to CRAN after doing a
little more work on the package.

I will be very interested to know if this resolves your problems. If
it doesn't, perhaps you could share your data with me off list? I
haven't got an example where I can reproduce the problem and such
examples are crucial when tracking down such bugs.

Cheers,
Rune

On 20 May 2012 17:49, Malcolm Fairbrother <m.fairbrother at bristol.ac.uk> wrote:
> Dear list,
>
> I am making use of the "clmm" function from the "ordinal" package, for fitting mixed models with ordinal outcomes.
>
> When fitting quite a few models, I get the error:
>
> Error in if (maxGrad < rho$ctrl$gradTol) { :
>  missing value where TRUE/FALSE needed
> In addition: Warning message:
> In sqrt(phi2) : NaNs produced
>
> Can anyone help me to understand why this may be happening, and/or whether there's anything obvious I should be doing different? With some seemingly small/arbitrary specification changes--or when taking (some) random subsets of my dataset--I don't get this message. But I'd like to find a way of reducing the frequency with which I do get it.
>
> I don't think there's anything unusual about my call:
>
> mod1 <- clmm(as.ordered(y) ~ x + (1 | fac1) + (1 | fac2), data=dat, link="logit", threshold="symmetric")
>
> I'm running:
> R version 2.15.0 (2012-03-30)
> Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit)
>
> Many thanks to anyone who can assist,
>
> Malcolm
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models



-- 
Rune Haubo Bojesen Christensen

Ph.D. Student, M.Sc. Eng.
Phone: (+45) 45 25 33 63
Mobile: (+45) 30 26 45 54

DTU Informatics, Section for Statistics
Technical University of Denmark, Build. 305, Room 122,
DK-2800 Kgs. Lyngby, Denmark



More information about the R-sig-mixed-models mailing list