[R] lmer and negative binomial family

[Ben Bolker]

John Sorkin wrote:
> 
> Patrick,
> I am not sure about this but 
> (1) I think you need to load the lme4 package, not MASS
> (2) I don't think neg.bin is allowable. I suggest you load lme4 and then
> type ?family
> 
> John
> 
> 

This conversation would probably go better on
r-sig-mixed-models at r-project.org ...
In the meanwhile --
   for a long time (I thought always, but I guess that's not true) the
allowable
families have been hard-coded within lme4, for purposes of efficiency.
The bottom line is that you can only use the following families:

famNms <- c("binomial", "gaussian", "Gamma", "inverse.gaussian",
            "poisson", "quasibinomial", "quasipoisson", "quasi")

'quasipoisson' might be good, if you're willing to settle for a quasi-
answer,
although you should be careful with random effects variances -- in earlier
versions there are some problems with the scaling there

   If you really need NB I would suggest glmmADMB ...


-- 
View this message in context: http://www.nabble.com/lmer-and-negative-binomial-family-tp26112313p26115443.html
Sent from the R help mailing list archive at Nabble.com.