AAS Reflections

This post originally appeared on the Software Carpentry website.

We just finished* a workshop at the American Astronomical Society. I was lucky to recruit 3 instructors (in addition to myself) - Matt Davis, Erik Bray, and Phil Rosenfield. We also had Pauline Barmby volunteer as a helper (and on the fly instructor). While I wrote this blog post, you will see comments inserted by the instructors.

Pre-Workshop Meetings

We started meeting at the beginning of October. Our first meeting was attended by only instructors. Despite a laughable number of technical difficulties we were able to decide on our curriculum, schedule, and who was teaching what. We held two other shorter meetings which included helpers to work out logistics and answer any questions (mostly mine). It was really nice to see everyone's face prior to the meeting. Additionally, we decided to meet the night before the workshop for a little more face time. I also set up an etherpad that held notes from all of the meetings. I found this very useful - other instructors and helpers can comment if they did.

I asked all helpers if they would like to teach a 15 minutes sections. This is something I like to do at every workshop I lead. It gives helpers a chance to teach (if they are not familiar with it) and to get a sense of what it is like to be in front of the class: how fast to go, where do students get lost, what is it like to explain concepts. With a 15 minute piece of instruction time - even if the lesson is a disaster (which none of ours were) the workshop is relatively unaffected.

I included the helpers and instructors on all correspondence prior to the workshop so that they knew the information that was given to the students on the first day and also to give them a window into the lead instructor's duties prior to the workshop.

Workshop Set-up

We decided to use the set-up test script from here. I modified the first test script to test for specific versions of numpy, scipy, matplotlib, and astropy. I emailed the modified scripts to students prior to the lesson and a link to the instructions for running them. We were able to head off many installation issues. [I think this workshop had some of the fewest installation problems I've seen. --Erik]

We also had a number of students show up early the first day for installation help. It is possible that having the installation script alerted people to when they did not have a correct installation. A number of students who ran the scripts downloaded them from the instructions rather than from my email. As a result they installed a number of unnecessary packages. These students encountered issues installing EasyMercurial and mayavi. Trevor King and I are currently working on an issue to comment out the packages we don't regularly teach in the default installation script. We had a lot of custom installations which often led to small differences and errors during the lessons (especially in Python). We also had a request from a student to modify the installation test script to handle dev versions.

I also sent students the Etherpad link prior to the workshop. At Erik's suggestion I created a short link to both the course website and etherpad - this worked really well. [Is this mentioned in any of the workshop setup instructions? I think it should be, if not. --Erik]

At my last workshop I received a number of comments regarding using Nano as the text editor. Specifically students wanted to know why we asked everyone to download Nano, used Nano, and then told them that they should probably use something else. As a result we had everyone download Sublime Text in addition to Nano. We still used Nano for the teaching git. Students seemed happy to know that they had Sublime Text and I saw many students opening and using it although we never formally used it in our lessons. Although we didn't try to link it to the command line, we encountered no installation issue. [I have a certain fondness for nano, but when we rely on it I do worry that it gives the impression to shell beginners that this is what they have to do if they want to edit files at the command-line. Not sure what to do about that. --Erik]

We left 3 hours at the end of the day as flex time. We thought our schedule may be ambitious and wanted to leave time to either finish the lessons or allow the audience to ask questions. This wound up being a good call.

Teaching

During each lesson I assign a different person to take notes on the etherpad and answer any questions that arise. I try to make sure a different person is doing this each lesson. View our etherpad here. [I've had mixed success with etherpad use in workshops. I think this one was sort of in the middle but overall it was very good to have. People seemed to appreciate the notes in there, and it was also useful as a shared copy/paste. A few people asked good questions in the etherpad but not as many as I like to see. --Erik]

Day 1

On day 1 Matt Davis taught the shell in the morning. Astronomers generally have experience in the shell and as a result after a brief survey of the room Matt was able to start the shell at lesson 5 on loops. Even those students less familiar with the shell were able to follow along.

Erik spent most of the rest of the day teaching Python. This group was mostly made up of people who had programmed before in another language and some who knew Python already. As a result Erik used his own lesson for teaching. This went slower than anticipated and we didn't cover moving from the notebook to command line scripts by the end of the lesson. Over dinner that night we discussed how we were going to adjust the schedule of the next day to account for this. We decided to give Pauline a chance to instruct and she worked on the end of the lesson that Erik was planning on covering and presented that on day 2.

At my last 2 workshops we've decided to teach code review instead of testing. I believe that while many students walk away from our testing unit without knowing how to apply the testing to their work, code review is a valuable tool that can prevent mistakes and help build a community at the student's home institution. It is also very easy to implement and can easily be taught in 45 minutes to an hour. Matt mocked up a terribly written code and we asked students to look at it in groups. We then went through each group and asked for a comment. I really enjoyed this section and thought it present a nice opportunity for students to implement the best coding practices they had just learned and for us to discuss it with them using a real example. It's also a nice way to finish off the day, by providing learners with an active experience that's not incredibly mentally-taxing.

We asked for good and bad stickies at the end of the day and reviewed them over dinner.

Day 2

We started day 2 with Git. I don't like to teach git entirely from the command line. Instead I like to present what we are doing and the command on slides - going back to the command line to do an example of what I've just told students and then putting up a slide with a challenge problem on it. I used slides.com to create an HTML5 slide show that I could version control, send students links to, and present to the class. My presentations can be viewed here: local git, remote git. Overall this strategy worked well. I followed the SWC lessons pretty closely for local git, although I left out the discussion of the gitignore file. I started out following the remote git lesson - having each student create a remote repository. However, to give them a sense of actually collaborating I had each person find a pair and assign one person to be Dracula and one person to be Wolfman. Wolfman then cloned Dracula's repository and I led them through a series of exercises where they first just pulled, modified, and pushed without the other person making any changes. Next they both modified but without conflicts. Finally they modified the same line and resolved conflicts. The whole Git lesson went much faster than I anticipated (I expected at least 3 hours and finished in 2.5).

Next Pauline presented her Python lesson. This involved walking through steps 1-4 of Erik's 'simcluster' notebook which shows how to run python code from the command line, how to parse command-line arguments using argparse, and how to break code up into functions. This went relatively smoothly, although Pauline suspects it might have been a bit too fast for everyone to follow.

In the afternoon Phil presented a short introduction to of matplotlib and Erik showed the learners around AstroPy. Most of the Python we did on day 2 was presented in a show-and-tell manner where students executed pre-written code rather than typed it themselves. This was an effective way to show them what is available. While this is good for future reference, I worry that we may be giving them more information than they can digest in a day.

One of the most consistent bits of feedback we got was that people really like doing group exercises like the code review exercise on day 1. Our Python lessons had little in the way of exercises, especially on day 2, but with these workshops there is always a balancing act between having exercises and trying to cover lots of material. (I'd like to experiment with multiple choice questions and peer instructions as a way to intersperse quick group discussions in workshops. -MD)

Overall - this workshop went very smoothly - no major issues. Thanks to all the instructors and helpers!

N.B. Instructor Interaction

A number of events have culminated in me thinking about interacting with my fellow instructors in front of the class.

  • I raised my hand a few times during Erik's Python lesson on day 1 and when called on asked if he wanted to explain something in a little more detail. I stayed at the back of the room while speaking. At different times Erik chose to either explain, say that was beyond the scope of the lesson, or that we would come back to it later.

  • We had a sticky note on the first day saying: "Erik should have listened to Azalee". I worried that my interruptions had been disrespectful to Erik in the eyes of the students and possibly reduced his effectiveness as an instructor. [I didn't feel that way, but I still need to work on my engangement with the room in general--it's hard to read them at my current level. --Erik]

  • In the Train the Trainers event at UC Davis Greg said that you should not interrupt fellow instructors as it splits the focus of students (which is already split between the instructor and the screen).

  • There was a discussion of this at our SWC workshop debriefing. Here Greg also mentioned that if you are going to interrupt you can walk towards the front of the classroom to signal that you want to say something and at that point the focus is less split.

I believe my take away from all of this is the following:

  • Most instructor to instructor interaction in front of the class is good (this is not scientific just anecdotal). It can draw the instructor's attention to something they left out or add information that benefits the class.

  • It is really important how this interaction comes about. You should never speak over your fellow instructor. I think ideally you can slowly walk towards the front with your hand up and hang on the sidelines until the instructor at the front of the room acknowledges you. This keeps the control of the lesson in the hands of the instructor who is leading the lesson.

  • Interruptions should be used sparingly. They should only clarify material never complicate it. If at all possible you should phrase your information in such a way that the instructor who is leading the discussion can explain it to the class. This allows the instructor to maintain the focus of the class and can continue to demonstrate their knowledge of the material that they are teaching.

  • This should be discussed before the workshop. Instructors should ask each other - are you ok with a interruption and how can I let you know that I have something to add?

  • Discussion beforehand also opens the door to an instructor saying - please don't interrupt - or to say - let me see how it goes - I will let you know if it's not working.

* This was written in January, its taken me a while to post it.

Dialogue & Discussion

Comments must follow our Code of Conduct.

Edit this page on Github