|
Abstract:
|
Hierarchical models provide a powerful framework for modeling ecological processes, as they allow for the natural modeling of conditional dependence. The use of hierarchical models in Ecology has increased dramatically in recent decades, due mostly to the ease with which many hierarchical models can be fit to data using Markov chain Monte Carlo (MCMC) algorithms. However, as data sizes increase and hierarchical models become more complex, MCMC algorithms can struggle to explore the full parameter space implied by a hierarchical model. We provide three examples of situations where a careful model choice and modern computing can lead to marginalization over, or removal of, latent variables in a hierarchical model.
|
