Our New Instructor Pipeline

This post originally appeared on the Software Carpentry website.

Last fall, we decided to reboot our instructor training course. We’ve tried a lot of things since then, and one of the biggest successes has been our new checkout procedure. In brief:

  1. After completing the instructor training course, the trainee picks one of the core lessons from each of the Carpentries she wants to teach and submit a new exercise for that lesson.

  2. She then takes part in an hour-long group discussion of that lesson led by an experienced instructor. She is expected to have gone through the lesson before this session so that she can ask lots of pointed questions during that hour. If the mentor leading the session feels that she is unprepared, she may be asked to do some more work and try again.

  3. She then does a demonstration lesson via screen sharing. In it, she is asked to teach a five-minute segment of her lesson chosen by the person running the session. Since she does not know in advance which five-minute segment she will be asked to teach, she must be ready to teach any part of the lesson. If the examiner feels that she needs to do more work, she will be given feedback and asked to try again, but if all goes well, she will get her badge. (Note that people don’t have to qualify separately for different topics: if you show that you’ve mastered one, we’ll trust that you’ll master others as needed.)

A dozen experienced instructors have run discussion sessions so far, and feedback has been very positive—everyone (both leaders and trainees) have found the sessions really useful. What’s more, trainers other than myself are now running the final demonstration lessons and deciding whether people are ready to teach for us. By mid-year, our rule will be that trainees are always examined by someone other than the person who ran their training course, for the same reason that PhD committees usually include external examiners.

It’s starting to look like a sustainable, scalable process, but there’s still lots of work to do:

  1. We need to do a better job of telling trainees and discussion leaders what’s expected of them. For example, trainees need to know that they’re responsible for mastering the whole of their chosen lesson, and that they need to check out separately for Data Carpentry and Software Carpentry.

  2. We need to automate scheduling and signup for discussion sessions and checkouts instead of using a pile of Etherpads and a flurry of emails. I expect we will do this via AMY, though I’m still leery of opening it up to hundreds of people.

  3. We need to make our expectations of trainees clearer. If someone signs up for a session, doesn’t show, then shows up late for another session they haven’t signed up for and asks the leader to wait while they read through the lesson, the discussion leader should know to hand them back to a trainer for a full and frank discussion of our reputation and their reliability.

  4. Equally, we need to cut infrequently taught material from our lessons so that trainees know what to focus on. For example, the material on building R packages is too advanced for most novice workshops; we should either move that material to a separate advanced lesson or mark it somehow.

  5. We need to set a time limit on completion, let everyone know what it is, and enforce it. Our current thought is to give trainees 90 days to wrap up, while being generous with waivers for extenuating circumstances.

  6. We need to follow up with the teams who took part in December 2015’s training and make sure they run the workshops that they promised to.

I’m sure lots of other things will come up, but we’re making progress. And it really is “we”: Steve Crouch, Christina Koch, Aleksandra Pawlik, and Tracy Teal are now certified instructor trainers, Ariel Rokem is in training to become one, and we are about add six more: Neal Davis, Rayna Harris, Lex Nederbragt, Anelda van der Walt, Belinda Weaver, and Jason Williams. By August, we may finally have the capacity to help all the people who come to us wanting to help their colleagues.

Dialogue & Discussion

Comments must follow our Code of Conduct.

Edit this page on Github