Hobbs awarded NSF grant to support Bayesian modeling short courses
By Caitlin Charlton
Professor Tom Hobbs from the Natural Resource Ecology Laboratory and Department of Ecosystem Science and Sustainability was recently awarded an NSF grant to teach short courses on the fundamental principles of Bayesian modeling and their application to ecological research.
Hobbs will be teaching these intensive two-week courses annually for three years in conjunction with statisticians Bailey Fosdick and Mevin Hooten beginning early summer 2022.
The courses aim to foster a collaborative environment for ecologists to learn how to develop compelling lines of inference within the Bayesian framework where they understand every single step of the process.
“What we are really trying to do is give people the gift and the power of understanding,” Hobbs said.
Hobbs has been teaching Bayesian modeling workshops around the world for the past 15 years. Trained as an ecologist, Hobbs said he is able to understand where intellectual hurdles arise for ecologists in Bayesian modeling and help accordingly.
“I know where you are going to have trouble understanding this, because I did,” Hobbs said.
Hobbs’ ecological expertise leading the room combined with the statistical knowledge of Fosdick and Hooten creates a well-rounded team to train ecologists in Bayesian modeling from an ecological perspective.
‘The essence of scientific work’
The driving objective of the course is to use sets of observations and models to answer what Hobbs said he thinks is the fundamental question of science.
“What is the probability that I would observe the data if my model is a faithful representation of the processes that gave rise to the data?” Hobbs said. “That is the essence of scientific work.”
The Bayesian framework allows that question to be answered effectively and accurately, Hobbs said. In the last series of workshops Hobbs led, there were over 20 publications acknowledging the workshop’s value and enhancement to the authors’ research.
Another goal of the course is interdisciplinary collaboration. Hobbs aims to teach participants a deep enough understanding of the fundamental principles of Bayesian models to allow them to collaborate with statisticians effectively.
“We are trying to make [participants] into people who can collaborate with statisticians, who will see [participants] as a resource,” Hobbs said.
The application process is competitive. Out of the 125 well qualified applicants that may apply, only 25 will be selected each year.
The course’s target audience ranges from advanced graduate students to post-docs to early-mid career researchers. Accessibility will be fostered by fully funding all participants and prioritizing advertisement at minority-serving institutions.
Now under a year away, Hobbs is excited to be able to share his knowledge and expertise with a diverse group of ecologists.
“The thing I am most excited about providing is the experience of learning things that has been the most intellectually satisfying in my whole career,” Hobbs said. “This stuff fits together beautifully; it is immensely powerful.”