StanCon 2018

The following guest post about StanCon2018 was written by Breck Baldwin of Stan.

Name that distribution. Mens t-shirts in the back, womens in the front.

Scholarships

We elected to extend our scholarship program to take all applicants who could manage the cost of travel to Asilomar while we covered the cost of lodging and meals. This turned out to be a very positive decision and it delivered a fun, knowledge-seeking audience that will carry Stan into the future. We encourage continuing to offer numerous scholarships in future StanCons. In the end we provided 50 scholarships which ended up perfectly offsetting registrations (80) and invited speakers (5) for a total of 135 which was our contracted minimum. Local attendees got us to 172 total.

Talks

We had 6 invited talks (40 + 10 min questions) who received room and board. One speaker dropped out and Andrew Gelman stepped up for a remote talk. We got 15 submitted talks (15 + 5 min questions) and all were accepted pending revisions. One submitted talk was retracted, we added two developer talks and had a “Mystery Talk” done by Aki for 17 total.

Wearable Poster Session

We modified the standard boring poster session into a “wear your poster” session at the first day reception and it was pretty much a hit. We had rolling poster acceptances until the day before the conference to help support people that could not attend without a presentation of some sort because of employer, school or other policy.

It was a hoot.

Tutorials

In an effort to maximize the attractiveness of StanCon we elected to have tutorials in the mornings and one afternoon session. We did not want to force people to choose between tutorials so we only allowed parallel sessions between the Intro to Stan tutorial and advanced tutorials. We surveyed attendees for tutorial interest along with t-shirt sizes and other preferences and saw tremendous interest in tutorials, with 60 people interested in Intro to Stan, 79 people interested in Advanced Hierarchical Models and so on. We thought these numbers were inflated but we were wrong—if anything we got more people than expected. In a mid-conference survey (Thursday onwards) it was clear that tutorials were the major driver for attending StanCon, assuming you believe that statisticians can properly read a double negative question.

The importance of tutorials was a big surprise but made sense given that 80 people were making a 7am tutorial (Thurs and Fri), and likely missing breakfast for the 8am to 10am tutorials as well. This is an untapped and future direction for Stan teaching. We are not serving the intermediate/advanced Stan modeler well.

Code of Conduct

As of writing we have no code of conduct issues. We were very proactive in communicating that inappropriate behavior was not welcome. We put the code of conduct with reporting options in each registration tote on yellow paper. A standard joke that began circulating was that some behavior was a “yellow sheet violation.” We hope that awareness shut down any possibility of problems.

Venue

Asilomar is beautiful, ideal and we had a lovely “summer camp vibe” to the entire event. On Thursday we had 14 hours of activities from 7am to 9pm. This cannot be replicated in a more urban environment and it significantly increased opportunities to network, learn about Stan and just have fun.

Asilomar is a special place. StanCon2018 Helsinki will be a more traditional urban/academic conference that will feel very differently from 3 meals a day with fellow sufferers of a sadistic 7am start time ending with blind wine tastings and air powered rockets.