2015 Post-workshop Instructor Debriefing, Round 2

This post originally appeared on the Software Carpentry website.

Greg Wilson and I recently hosted a second post-workshop debriefing session for January to capture more experiences and lessons learned from instructors in the field. This meeting was attended by 10 instructors covering four recent workshops. We discussed how the workshops went, what worked, what didn't, and what could be improved.

There is always something new to learn from instructors' experiences, particularly from this meeting, as we had the chance to discuss a Data Carpentry workshop at Cornell University as well as a three-day "Beyond the Code" workshop at the NERC/University of Leeds. Some highlights are discussed below. Please see the etherpad for detailed notes.

Instruction Methods and Technology

Dual screens were available at the University of Washington and Oxford workshops, which was reported to have worked very well. We've discussed before that it is helpful for the instructor to use a tablet (or second computer) to consult the lessons while live coding. Having two projectors seems to be even better by making it possible for the participants to refer to the lessons material while also keeping the code visible.
Some take-home lessons were that live coding on screen is much better received than prepared slide decks. Also, you can't have too many hands-on exercises, which are popular with participants.
Another common theme from this retrospective was that it was difficult to estimate and/or stick to lesson timing, which is probably related in part to new or newly refactored lesson materials.
We've noticed in workshops that adoption and use of the etherpad can be somewhat spotty, not gaining traction or trailing off as the workshop proceeds. When the etherpad is actively used by the participants, it can be a very effective teaching adjunct and resource for real-time discussions, answers to common questions and helpful links. Almost all participants will be unfamiliar with the etherpad at the start or may not immediately see its value but, when it works, it works well and builds momentum. More active participants can be encouraged to post content as well to reduce the burden on the instructors and helpers.

Data Carpentry Workshop at Cornell

This workshop was run by Erika Mudrak (CSCU), Jeramia Ory (Kings College), and Emily Davenport (Cornell University).

Covered Excel, Open Refine, SQL, R, and shell.
OpenRefine, SQL and R were all very well received.
Some technical issues.
Some students were extremely under-prepared to use their own computer.
Excel consumed a disproportionate amount of time.
Most people were at a good level to learn and got something out of each part.
Need to work on motivating participants to use each tool by introducing why it is helpful for them (but not all users have to use them all).
A few personalities in the front of the room dominated the mood (also the under-prepared ones).

Software Carpentry Workshop at the University of Leeds

This workshop was run by Martin Callaghan, Aleksandra Pawlik, Andrew Walker, Aaron O'Leary and Peter Willetts. It was the second iteration of the "Beyond the Code" format workshops held at Leeds, sponsosored by the Natural Environment Research Council (NERC). The first two days are based on a modified Software Carpentry workshop model. The third day of these workshops was added to allow small groups of students to work together to practice their skills and begin to develop software that may be useful for their research projects.

See Andrew Walker's post on the Software Carpentry blog for a detailed description of the first such workshop.

Good things: the "day three mini-hacking" sessions seemed to work well. Git went down well with the participants.
Problems: some minor time-keeping issues (probably worth having a plan for how to signal the teaching instructor during the lesson that time is passing), problems with etherpad (but less of a problem than in November), didn't always include enough hands-on during taught bits (setup of SWC lessons not great for this - challenges tend to be at the bottom). Fairly quiet group.
Split into number of short sessions, opportunity for attendees to work at their own pace and lots of helpers and decent food and coffee!
Superb organization- see Andrew above.
Would like to develop this towards a more problem-solving approach and applying some flipped-teaching principles to allow people to work at their own pace.
Pre-planning was essential. We spend quite some time agreeing content and format, deciding who was goiung to do what and how and produced quite a detailed delivery plan (not that we followed it too closely...)
Tried to make the Python materials a bit more bespoke to be relevant to our attendees. For example, I did a session on creating maps using Basemap and plotting data onto them. For one session, I used a very short set of slides to introduce session objectives and to summarize.
We followed the material quite closely for Bash, went a bit beyond the Git material.