Closed osupplisson closed 3 months ago
Hi, this should now be fixed in development version 2.10.1.9001. Also note a small correction to your component definition:
cmp <- ~intercept(1) + z(z,model = "linear",mean.linear = 0,prec.linear = 1)
lik1 <- like(
data = data,
Ntrials = Ntrials,
###Should be a likelihood arguement
scale = s,
family = "betabinomialna",
formula = y ~ intercept + z
)
Hi,
I was trying to fit a Gaussian approximation to the betabinomial model using the 'betabinomialna' likelihood from INLA (https://inla.r-inla-download.org/r-inla.org/doc/likelihood/betabinomialna.pdf).
It appeared that the
scale
argument from theinla()
call is currently not a valid option for thelike()
function. I believe it should be the case as this is a parameter that can change across likelihood (like E and N) ?Best, Olivier
(ps: I think, the same thing holds for the
strata
argument)Please, find a small reprex hereafter: