Arnaud - 2013-08-06

Hi all,

Using a hierarchical model, I am trying to fit observed data to predicted data obtained through the population dynamic part.

One of my output is the number of individual dead each year and I want to link this number to the number of dead individual observed each year considering that only a limited proportion can be recovered.

Thus, I considered that observed individuals are the result of a binomial trial using the following code.

for (i in 1:Nyears){
    deadObs[i] ~ dbin(pObs, round(deadPred[i]))

I first used a uniform prior for pObs, but now I want to include an error on this parameter.

  • The use of a prior taking the form of a beta distribution seems correct, however, I do not see how to include an error that I can control and make "equivalent" among years.

  • Another thing is that observing a dead animal is like taking a ball from an urn without replacement ... so I wonder if I need to use a hypergeometric distribution in place of the binomial.

Any suggestion would be greatly appreciated