Hierarchical models

We are now familiar with the basic concepts of statistical inference and the two philosophies that are commonly adopted to make the inferential statements. In this lecture, we will look at making inferential statements about realistic and hence complex ecological models. In the rest of the course, we will write the description of the model but will not discuss the philosophical aspects in detail. We will mostly use a graphical model and its JAGS version. We will provide tools to obtain either Bayesian or Frequentist inferential statements. We will discuss pros and cons of these inferential statements. The choice of the inferential statement will be left to the scientist.

We start with simulation so that we can compare the results to the true parameter values (which is never the case with real data). Then we show the Bayesian implementation, and how to modify that for data cloning based frequentis inference. We also showcase peculiar properties of these model classes to highlight similarities and differences of the different approaches, and to demonstrate the potential pitfalls and how to avoid them.