[R] negative binomial lmer

Ben Bolker bolker at ufl.edu
Sat Dec 13 18:18:07 CET 2008

Marty Kardos wrote:
> Hi;
> I am running generalized linear mixed models (GLMMs) with the lmer
> function
> from the lme4 package in R 2.6.2. My response variable is overdispersed,
> and
> I would like (if possible) to run a negative binomial GLMM with lmer if
> possible.  I saw a posting from November 15, 2007 which indicated that
> there
> was a way to get lmer to work with negative binomial by assigning: family
> =
> negative.binomial(theta=2.5) (or whatever value I would like to specify
> for
> theta).  However I have not been able to get lmer to work while attempting
> to specify the family as negative.binomial.  I get an error message that
> says "could not find fuction "negative.binomial".
> Is there a way to make lmer run negative binomial GLMMs with lmer in the
> current version of R?
> Thanks so much!
> Marty

Alas, this is simply not possible in the current version of lmer, and is not
completely trivial to fix (i.e., you won't be able to hack a couple of lines
in the R code and make it work).  If you really need a negative binomial
GLMM then the glmmADMB package is probably your best bet.

  There's an active thread on the topic of overdispersion and GLMMs
(now diverging to more general philosophy of statistics) running on 
the r-sig-mixed-models mailing list; I'd suggest you send follow-up 
questions to that forum.

    Ben Bolker

View this message in context: http://www.nabble.com/negative-binomial-lmer-tp20966631p20992317.html
Sent from the R help mailing list archive at Nabble.com.

More information about the R-help mailing list