(Beginner question) Minimum no. sites to fit a Bernoulli GLM with spatial correlation

[Berin]

I’ve been taught the rule of thumb that a minimum of 50 sampling locations is required to a fit a model with spatial correlation in INLA. In the case of Bernoulli models (e.g., modelling binary response variables such as survival, or modelling the binary part of ZAG/hurdle model), does the minimum 50 locations apply to the total number of Bernoulli observations, or is it the number of locations with ‘successes’ or ‘failures’ (whichever is less common of the two)? Thank you.

[Håvard Rue]

Never heared of that rule of thumb, but of course, the information you have in
the data about estimating the spatial correlation, do depends also on the number
of sampling locations (and also where they are located). binary data is also
less informative. it is possible to set, a spatial range to mimic what you think
its there, then it becomes more like a scenario model, where, _if_ the range is
this value, _then_ the results are like this. you can have different scenarios.

> --
> 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/af55e8f0-56ee-43d3-a200-d04137567b3fn%40googlegroups.com
> .

--
Håvard Rue
hr...@r-inla.org