Greetings,
I would like to use WinBUGS to simulate from the prior
predictive distribution from a hierarchical model (i.e. simulate
using only the priors; no data). My unsuccessful approach was just to
take an old program with a working model with covariates, and the priors,
and hyperpriors specified, and removed the the response variable,
y).
Winbugs allows the model, the only comment I get is that all the
nodes are not initialized so I click "gen inits" which generates initial
values so thats seems fine.
But when you actually try to run the samples I get a trap.
The error at the top of the trap is "undefined real result".
I would very much appreciate suggestions on how I could
simulate from the prior predictive distribution.
Thank you,
Suzette
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|