Awardees: Xiao-Li Meng (FAS), Joseph Blitzstein (FAS), David Harrington (SPH/FAS)

Summary: Awardees plan to organize a two-day workshop that brings together faculty and students from across the University to discuss pedagogical strategies for teaching statistics.

“One of the things that inspired me was the original meeting of the HILT symposium, and how much was learned from getting people with experience in the classroom into the same room to talk about the problems they were facing. It struck us that it would be great to start that way.” —David Harrington

Amazed to see that there are so many people at Harvard teaching statistics

Awardees organized a two-day workshop held to bring together faculty and instructors from across the university to discuss pedagogical strategies for teaching statistics. The workshop—dubbed CASTLE: Coordinating and Advancing Statistical Teaching and Learning—took place on April 11-12, 2014 in the Lamont Forum Room. Around 30 participants engaged in presentations and conversation about the teaching of statistics and machine learning, pedagogical innovations, favorite examples, common misconceptions and roadblocks, TF training, online education, and assessment.

Joe Blitzstein, one of the event organizers, said multiple participants expressed that they were amazed to see how many people at Harvard are teaching statistics since they rarely cross paths in their day-to-day activities. “There are so many statistics courses taught in the various schools at Harvard, but so little communication and coordination for sharing new ideas, data, and resources.”

Blitzstein and his collaborators deliberately set an informal agenda structure, inviting everyone to present their teaching and methodology, but allowing lots of time for Q&A and discussion. Those who taught introductory courses presented earlier in the workshop, followed by more advanced courses. Participants found that many similar concepts emerged, but that they learned different applications from one another.

As a result of this workshop, there is new enthusiasm from participants and observers to actively coordinate statistical activities on campus. The group has plans to develop a website for the statistics community at Harvard, including shared syllabi, contact information, and plans for future workshops. Blitzstein hopes that the development of this website resource will streamline the logistical planning for future events, at the very least to ensure the complete invitation list as compiling the list for the inaugural event proved challenging. In addition, such a resource would make it easier for students to determine the target audience of a particular course, where it falls on the theoretical-applied spectrum, how courses are arranged within a department as well as how they compare between departments.

Moving forward, the workshop will likely be offered as an annual one-day event. Blitzstein said it is possible that future iterations will include themed sections in addition to pedagogy, such as where to get data and how to clean a data set. He also emphasized that students benefit from this increased exposure to the greater statistics community on campus, as participating instructors will be more informed academic advisors. For example, as co-directors of Undergraduate Studies in the Statistics Department, Blitzstein and David Harrington developed a family tree graphic to help students understand the various offerings and paths, a model that might be useful for other departments.