AFS Continuing Education
ROOM: RSCC, A17
(*Bayesian courses may be taken separately or together as a two-day series)
The benefits of undertaking an analysis using Bayesian rather than classical methods become far more apparent for complex models. These include models with non-normal variance structures, multiple sources of variability (i.e., hierarchical models), and especially those that attempt to estimate characteristics of both latent and observed processes from the data. This workshop, designed for participants with an introductory understanding of BUGS and Bayesian concepts, will extend the material covered in the introductory first day to illustrate ways to accommodate these complexities in JAGS models. Again, application rather than theory will take precedence as participants learn about generalized linear models, checks for model adequacy, mixed effect models, and state-space models and how to code them in JAGS. Specifically, participants will explore the implementation and interpretation of logistic, Poisson, and negative-binomial regression models, the inclusion of random slopes and/or intercepts in these models, and the basic state-space Cormack-Jolly-Seber model with extensions to accommodate event- or individual-specific effects on survival or detection. Depending on time and participant interest, other state-space models may be demonstrated including a biomass dynamic model and a spawner-recruit analysis.