A New Paradigm for Statistical Learning at Conferences
Methods Incubator for Junior Researchers

What are the best things about statistics conferences?

  • Connecting with colleagues
  • Meeting new people and learning about their research
  • Finding inspiration in new ideas
  • Sharing your work with an engaged audience

But how often do these really happen at meetings?

Last summer, fresh from the Joint Statistical Meetings, we (Sherri Rose and Laura Hatfield) wondered aloud how we could shake up statistics conferences, especially for early-career statisticians. In our experience, major conferences focus on passive participation, offering a panoply of parallel sessions jammed with 20-minute talks. Junior researchers do not give many high-profile invited talks, and the audiences for contributed sessions are usually…thin.

The question is how to 'disrupt' conferences, to add interactive opportunities with tangible benefits for junior researchers. Imagine conference activities that emphasize communicating with broader audiences, spreading innovation across subfields, and fostering engagement with works in progress. Thus, we conceived a novel "Methods Incubator" workshop, debuting at the International Conference on Health Policy Statistics (ICHPS) on October 7, 2015. 

The premise is fairly simple. We've recruited six "team leaders" to propose interesting methods problems in varied areas. Early-career researchers (e.g., postdocs, junior faculty and equivalent in government and industry) indicate their interest by submitting a paragraph on how they might solve a particular team's problem and their CV to us. Team leaders will then invite 6-8 applicants to join each team. Before the conference, invited team members will submit a 1-2 page description of their approach to the problem, which could include preliminary data analysis of provided data. The idea is that each team member will arrive at the conference having thought critically about the problem and can spend the 4-hour workshop productively. At the workshop, everyone will briefly share their ideas within teams. Then teams will collaboratively work on the most promising ideas for 2-3 hours (likely each team will break into smaller project groups based on mutual interest in particular approaches). The aim is for project groups to make progress toward a manuscript at the conference and continue to collaborate long after.

The inaugural Methods Incubator for Junior Researchers at ICHPS features the following topics:

  • Designing behavior intervention trials for challenging circumstances (Team Leader: Michael Baiocchi, Stanford)
  • Statistical design of physician payment programs (Team Leader: Frank Yoon, Mathematica)
  • Nonparametric methods for difference-in-difference estimation (Team Leader: Sherri Rose, Harvard)
  • Combining health utilities across multiple outcomes (Team Leader: Laura Hatfield, Harvard)
  • Handling known and suspected missing data in event sampling studies (Team Leader: Stephanie Kovalchik, RAND)
  • Spatial confounding in generalized linear mixed models (Team Leader: John Hughes, University of Minnesota)

To learn more, please read the official ASA document on the Methods Incubator, which includes detailed descriptions of proposed topics.

And be sure to mark these key dates on your calendar:

  • June 30: Submit 1-2 paragraph description of your idea for a particular team's problem and your CV to Laura
  • July 31: Team leaders will invite participants and distribute registration forms
  • August 25: Workshop registration deadline
  • August 30: Submit 1-2 page problem approach description to the team leader
  • October 7: Workshop!

How neat would it be to attend a conference and leave with half a paper written with new collaborators?

If you have questions please feel free to email us.