0

I am trying two compare two hierarchical Bayesian models using the following output from R:

I have understood that the DIC is used to compare Bayesian hierarchical models and that the lowest its value, the better.

Basically, I'm having two models and these are the outputs from R:

Model 1: Mean Deviance: -11877 Penalty: 405.3 Penalized deviance: -11472

Model 2: Mean Deviance: -10828 Penalty: 60483 Penalized Deviance: 49655

What exactly do each of these terms mean? Can they be negative and is it a problem if they are negative? Is penalized deviance equal to DIC? Which of these two models do you believe is better from these two results?

Thank you!

jmemax
  • 1
  • 1
    I suggest you read the documentation that comes with this R package and the [Wikipedia page](https://en.wikipedia.org/wiki/Deviance_information_criterion) on the topic. – Xi'an May 05 '21 at 18:08

0 Answers0