Course | Linear Algebra for Artificial Intelligence and Informatics |
Lecturer | Leo Dorst |
Context | Basic linear algebra, from vectors to matrix decomposition (SVD) and least squares techniques for a large group of 400 students |
Challenge | Not enough time/opportunity to provide students with enough feedback on the basics |
Design | e-learning exercises about the basics with automated feedback |
Evaluation | Students kept up their basic skills and the TA’s could focus on the difficult parts |
Can you tell something about your course before the innovation?
The students needed to keep up with the fundamentals, in time for the next lecture, but we could only give them feedback on homework once a week, whereas the lectures were twice a week. Also, correcting the basics is not the best use of a TA’s time. And the students tended to plagiarize because of this.
What intervention was chosen?
Part of the homework, at the basic drill level, became SOWISO exercises to be completed before the next lecture. Some points could be earned by this.
Did it solve the issue?
Yes, students kept up their basic skills, TAs could focus on the next level in the werkcollege and the paper homework. And the randomization made it personal.
What was the students’ experience?
Students liked it, and the distribution of final grades became more of a bell-curve (around 7) whereas earlier it tended to be rather uniform.
Are you going to use it again? If yes, what would you change in the next iteration?
I already did. And I also tried one year to make part of the exam ‘SOWISO-corrected’, in order to correct it more quickly. This I will not repeat, but rather try using ANS – it was hard to ask reasonable exam questions in SOWISO.
How much (extra) work did it cost you? Does it outweigh the benefits?
I actually first did this in ONBETWIST, using mostly exercises that were already present there. Those were translated to SOWISO; the effort was not done by me. It was definitely worth it, and not only for me: now we have lots of LA drills in SOWISO.
Do you recommend this approach to other lecturers?
Yes, for subjects that are amenable to this. In fact, I already did, and Calculus and Statistics (by Homburg and van Es) followed the e-drill principle (directly to SOWISO rather than via ONBETWIST).
Is there anything else you’d like to add?
There was great supportive help from Henk Kuijpers (ONBETWIST), and from André Heck, Marthe Schut and Jolien Oomens (SOWISO) to enable both the e-drills and the trial with the exams. This was a big surprise, it is rare to have colleagues in such an innovation. It made this innovation much easier than I had ever thought it would; it should be better-known that you do not have to do these innovations by yourself.