variability/Reliability of INLA spatial estimates

[Alokesh Manna]

Hi INLA users, 

I see for the spatial coefficients INLA provides a sd with the mean of the coefficients. I used a besag model in Poisson regression. Is there any way to control the sd and make a more accurate mean or estimate?

I provide an ss for my computation.

Alokesh

[Håvard Rue]

The mean is already corrected for. What we see, by experience, is that the stdev
is pretty accurate, while we might lack a little skewness. However, the main
source for skewness in marginals in the latent, is due to integration of
hyperparameters. there are several papers studying accuracy, but for Poisson
counts it should be pretty pretty good. I would not worrry about it

> --
> 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/accd06d1-08ba-4cd5-ae21-b3ebaeeda395n%40googlegroups.com
> .

--
Håvard Rue
hr...@r-inla.org

[Alokesh Manna]

Thanks professor Rue!

What I observed is that for the same data if I run my model 3 times, there is a small variability. I wonder why this is the case (diff is the difference)? The estimated hyperparameters also fluctuate a bit. I provide estimates of the fluctuation of one hyperparameter here based on 100 simulations.

Alokesh