The National Science Foundation (award #: DEB-1145200) is sponsoring an
annual, two-week workshop to provide intensive training in Bayesian modeling for post doctoral researchers, academic faculty, and agency scientists. Applicants must have earned a PhD to be considered for the workshop, and applications will not be accepted from graduate students. Twenty participants will be invited each year. There will be no cost for participation in the workshop. A $1000 stipend will be provided to each U.S. participant, and a $700 stipend will be provided to each international participant (due to taxes withheld), to defray costs of travel. The fourth workshop will be held May 18-27, 2016 at Colorado State University in Fort Collins, CO.
Goals of the Workshop
1. Provide a principles-based understanding of Bayesian methods needed to train students, to evaluate papers and proposals, and to solve research problems.
2. Communicate the statistical concepts and vocabulary needed to foster collaboration between ecologists and statisticians.
3. Provide the conceptual foundations and quantitative confidence needed for self-teaching modern analytical methods.
At the end of each workshop participants will be able to:
1. Explain key principles of Bayesian statistics including the concepts of joint, conditional, and marginal probabilities, posterior and prior distributions, likelihood, conjugacy, conditioning, and the relationship among simple Bayesian, hierarchical Bayesian, and maximum likelihood methods.
2. Use basic statistical distributions (e.g., binomial, Poisson, normal, lognormal, multinomial, beta, Dirichlet, gamma) to write joint and conditional posterior distributions for Bayesian models.
3. Explain how Markov chain Monte Carlo (MCMC) methods can be used to estimate the posterior distributions of parameters. Write algorithms and computer code in R implementing MCMC methods to estimate parameters in simple models.
4. Use JAGS software to implement MCMC methods for estimating posterior distributions of parameters, latent states, and derived quantities. Evaluate model convergence. Assess goodness of fit of models to data.
5. Develop and implement hierarchical models that explicitly partition uncertainties.
6. Understand approaches for evaluating the strength of evidence in data for alternative models of ecological processes.
The course will include lectures and laboratory exercises. Labs will emphasize problem solving requiring programming in R and JAGS. There will be four to six group projects using data provided by participants. The projects will be aimed at producing published manuscripts.
Tom Hobbs has taught ecological modeling at Colorado State University for 12 years. His course has evolved over time; during the last six years it has emphasized likelihood-based and Bayesian methods for model-data assimilation (http://classes.warnercnr.colostate.edu/nr575/). He has also taught short courses for the U.S. Geological Survey, the Grimso Wildlife Research Institute and the Department of Ecology, Swedish Agricultural University. Hobbs takes special pride in making challenging, quantitative concepts clear and accessible to ecologists who never considered themselves to be particularly adept with mathematics and statistics.
Mevin Hooten currently holds joint appointments in the the Department of Fish, Wildlife, & Conservation Biology and Statistics Department at Colorado State University. He previously taught both Applied Spatial Statistics and Bayesian Statistics while on the faculty at Utah State University. He currently teaches a course entitled Hierarchical Modeling in Ecology. At the 2009 Annual Landscape Ecology Meeting (US-IALE) he delivered a well-attended short-course on Bayesian Methods for Landscape Ecologists. With a formal background in both ecological science (BS and MS) and statistics (PhD), he has a unique perspective on statistical pedagogy, having taught for two different audiences from two different home departments himself (i.e., Statistics and Fish, Wildlife, & Conservation Biology).
Kiona Ogle has degrees in mathematics, statistics, and biology, and has been formally teaching Bayesian and ecological modeling to peers, students, or post-docs since about 2001. While an assistant professor at the University of Wyoming (2006-2010), Ogle taught "Hierarchical Bayesian Modeling in Ecology," two semesters of "Ecological Systems Modeling," and three semesters of "Bayesian Data Analysis." The latter two courses involved lectures and a separate, weekly lab session that provided students with hands-on experience. Ogle has also served as the lead organizer for a daylong workshop on "A Brief Introduction to Hierarchical Bayesian Modeling in Ecology," which has been organized annually since 2006 for the Ecological Society of America (ESA) annual meetings. Ogle has also offered a similar workshop for the Spanish Ecological Society meeting, a more detailed version for colleagues at the Max Plank Institute for Biogeochemistry in Jena, and a truncated version for a group of biologists at Kansas State University.
Maria Uriarte teaches a statistical modeling course at Columbia University (EEEB G5010). In collaboration with Charles Canham, Uriarte has also taught a two-week workshop on the application of likelihood methods in forest ecology. The course has been taught four times in the United States, and on Uriarte's initiative, once in Chile, and once in Puerto Rico. The course drew, and continues to draw, students from a wide range of countries and interests (so far ~80 students from 11 nations).
Participants must have a general, working knowledge of R.
Send a two page curriculum vitae as well as a statement of interest in the course that speaks to the following evaluation criteria:
1. Opportunity to train others in principles and methods learned in the course through teaching or by mentoring research.
2. Opportunity to use principles and methods learned in the course to influence management and policy.
3. Interest in applying course material to their own research.
4. Need for principles based training in Bayesian modeling, and demonstrated lack of opportunities at his/her home institution.
In addition, applicants will be evaluated on their contribution to the diversity of participants, including diversity in sex, age, ethnicity, research interests, geography, and institution.
Applications must be received no later than January 15, 2016. Send one PDF file with the CV and qualifications letter as an electronic attachment to Jill Lackett. Please follow the following naming convention for the file: LastName_FirstName_Bayes_2016. Invitations to successful applicants will be sent by February 15, 2016.