This is the program for the 2010 Joint Statistical Meetings in Vancouver, British Columbia.
|CE_10C||Sun, 8/1/2010, 1:00 PM - 5:00 PM||CC-2&3 (East)|
|Bayesian Ecology: Hierarchical Modeling for Ecological Processes — Continuing Education Course|
|Section on Bayesian Statistical Science|
|Instructor(s): Alan E. Gelfand, Duke University, James Clark, Duke University|
|The past twenty years has seen an expanding Bayesian presence in ecological research. We see a paradigm shift that goes beyond the acceptance of Bayesian inference. Specifically, we see an increased role for graphical approaches for model building and associated computation. These graphical models are hierarchical models which we capture in the multi-level specification, f(data|process, parameters) f(process|parameters) f(parameters). The goal of this course is to illuminate the richness of this specification and its focus on learning about an ecological process by studying the posterior distribution for the process given the data. We move from designed data collection to observational data, from controlled experiments to an integrated process view. We do not ignore knowledge from earlier experiments. Rather, we incorporate all available information about the process including such experiments but also scientific (mechanistic and theoretical) process knowledge. The course will present, in substantial detail with data analysis, four illustrative process modeling contexts - forest dynamics, species distributions and biodiversity, pathogens on hosts, and species diffusion processes - to illustrate the above ideas. Pre-requisites are courses in mathematical statistics and in standard statistical modeling and inference. Familiarity with basics of Bayesian inference and MCMC methodology will be helpful.|