Request: "xnbinomial" likelihood for non-integer counts

27 views
Skip to first unread message

bolis...@gmail.com

unread,
Jul 28, 2026, 6:50:38 AM (13 days ago) Jul 28
to R-inla discussion group
Hi,

Similar to the implementation of xbinomial (see https://groups.google.com/g/r-inla-discussion-group/c/FLpdv5jB90w/m/kSwlgxb9AgAJ), would it be possible to add xnbinomial to support the use of non-integer counts for survey-weighted data?

Many thanks,
Oli.

Helpdesk (Haavard Rue)

unread,
Jul 28, 2026, 1:20:41 PM (13 days ago) Jul 28
to bolis...@gmail.com, R-inla discussion group

that is doable. I'm back at KAUST next week.

bare in mind, that the 'size' or 'dispersion' parameter, kind of looses its
meaning since the likelihood is no longer normalized for continous response. so
I do not know what `it is' anymore.

with this, do you know what benefite that neg.binom has compared to xpoisson ?
if you add a iid effect using poisson, you essentially get back the
neg.binomial, can you use the same trick for the 'xpoisson' to mimic xnbinom ?

Best
H
> --
> You received this message because you are subscribed to the Google Groups "R-
> inla discussion group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to r-inla-discussion...@googlegroups.com.
> To view this discussion, visit
> https://groups.google.com/d/msgid/r-inla-discussion-group/63a69a98-620f-44ef-94b8-7797a4347d7bn%40googlegroups.com
> .

--
Håvard Rue
he...@r-inla.org
Reply all
Reply to author
Forward
0 new messages