Fluid dynamics course, 2020
Now that the semester is over, I can take a few moments to reflect over the 2020 edition of my fluid dynamics course at the University Otto von Guericke of Magdeburg.
The semester began right as Germany shut down due to the pandemic crisis. The course, with 153 students, took place entirely online. I spent nearly half of the semester working from abroad, attending to a close relative. I was never able to meet any student physically.
The course progressed on a weekly basis. I released one chapter every week, a single PDF document containing lecture notes and problems. In the document, I linked to YouTube videos (on average, five per chapter) where I covered important elements of the theory, and solved problems step-by-step. I also linked to interesting illustrative videos. In weekly, two-hour Zoom sessions, I answered any and all questions that were thrown my way.
In addition, every student received individualized homework as part of the peer-graded homework program I developed with my colleagues Mathias Magdowski, Germán Santa-Maria and Jochen König. This year, participation in the homework was mandatory and counted 40% towards the final course grade. I hoped to run four instances of the homework, but in the end, because of personal matters, only three pieces of homework were handed out.
A series of weekly four-question quizzes, run through the University’s Moodle instance, added up to 10% of the final grade. The remaining 50% of the final grade consisted of the final two-hour invigilated examination, with the same structure (choose three problems among five) as in former years.
A lot has happened through the semester, and I have many mixed feelings about this edition.
First, I am glad I learned to record videos, and I think I got it right. A good-quality, subtitled recording of a problem being solved by hand (I used a Wacom tablet, an external microphone and an external webcam plugged into a laptop) is a much better tool to learn with than a live session in a classroom with a chalkboard. I chose YouTube as a compromise, because it suits students well, but will try to migrate to a PeerTube instance in the future.
Administrative management was a nightmare and cost me hours of work weekly throughout the semester. My university does not communicate well with either teachers or students —the local TV station knew of the revised semester dates before I did— and predictably the communication between teachers and students is chaotic. Additionally, my university has no effective control over the course registration process: for example, just getting an up-to-date list of student emails is a challenge. I hope better management will alleviate me of that load in the future.
Communication channels between students remained out of my reach for the complete semester. I heard WhatsApp groups are the thing; in any case the Moodle discussion boards of my course failed to reach critical mass. Too many students address me directly. I tried to redirect requests to the interactive Zoom sessions and Moodle forums, but in the end I count only 174 Moodle posts, and 860 emails sent and received over the semester (I do not count the 1800 mails processed automatically as part of the peer-graded homework program). An average of 4.5 mails received, 4.5 sent every day is just too much for one class. A few students are disrespectful (although always reverent), and many have completely unrealistic expectations, not having truly realized what it means to be a single instructor with 153 students on the side of research-related duties.
Zoom sessions went very well. The technical robustness of Zoom impressed me, especially in view of the low level of computer literacy and equipment of my students. Attendance (share of exam-goers present for 45 minutes or more) began at 65%, decreasing linearly with time down to 17% in the last session. I regret that so few students chose to turn on their camera, even though I respect their decision. As a teacher, I yearn for visual contact and body language during interactive sessions, but this semester, I ended up staring mostly at a wall of usernames.
One nice success was introducing illustrations of people in the lecture notes, to pass across important points in paragraphs titled “advice from an expert”. I had some illustrations made to order, and licensed others under a limited-use, commercial license, all following the advice of my colleague Dr. Kristin Hecht.
Those paragraphs had two purposes. First and foremost, I am tired of fluid mechanics being narrated mostly by old white men, and wanted to put different-looking people in charge of giving technical advice to students. I hope my students, almost none of whom look like me, can identify themselves with some of the persona displayed in the lecture notes now. Second, using this technique helped me with the tone of the lecture notes, which I had long found to be slightly boring and too serious. Now, I can keep the technical content short and precise, while indulging in a more relaxed tone on occasion, addressing the reader directly. For good measure, I also sprinkled links to a few of my favorite science webcomics (such as xkcd) across the chapters.
The largest pain point remains the grading. While the distribution of final grades remains largely the same as in the last five years, there is a large discrepancy between coursework and exam grades.
In the peer-graded homework program, there is evidence that a sizable share of students participated in bad faith, which resulted in low-quality grading of other students. This is an inherent weakness of the program: if too many students, even when handed a complete step-by-step worked-out guide, are unwilling or unable to identify the source of an error in the papers they grade, then they cannot be easily “sorted out” of the system, and global student motivation and learning will suffer.
There is also plenty of evidence that the system of weekly quizzes was abused. I had thought that because the stakes were low (quizzes amount to 10% of the final grade, so any one question weighs only 0,25%), there would be no incentive for gaming, and I was wrong.
Because of those issues, I went into the exam with the impression that coursework grades did not reflect the true level of learning of a significant number of students. The exam proved me right: with 55% of participants failing the exam proper, and 33% not even clearing a quarter of points, it was the worst student performance in record. This necessarily brings into question whether deploying coursework is actually worth the effort, since this year, it objectively correlates with lower learning.
There are many factors that come into play here, not all of which are under my control. First, this year the University lifted restrictions on the number of attempts allowed for each exam, and so some students may have been temped to have a “free try” (passed exams cannot be repeated however, which makes for weird dynamics). Second, grades secured with coursework reduce the pressure exerted on students during the exam. I have long understood that grades act as a metric for the student’s time budgeting, both within my course and in between different courses. This is completely natural, and I wouldn’t blame students for putting focus on other exams where the stakes were higher. (Proper quality measurement and management would substantiate those statements and help with strategizing, but unfortunately, my university is not there yet either.) And finally, the pandemic disproportionately affected students in my course, most of whom come from abroad. I can’t imagine what it means to work on an exam when you are barred from flying back home to take care of loved ones, or uncertain whether you can even keep studying through the coming semester.
Looking ahead, I need to make the grades better reflect the level of learning. Quizzes won’t be graded anymore, and I will likely make homework an optional, low-weight program again.
Ultimately, it is clear that there are students in my course who have not attended a single Zoom session during the semester. There are students who do not think it useful to invest 20 minutes watching the exam prep video I sent them before making an attempt. There are students who are not able to manipulate a simple equation that contains only symbols, and there are students who have no inhibition against writing uninterrupted nonsense for two hours on a paper. Surely, any of those last two issues preclude any success in an engineering fluid dynamics course. Before I make coursework a large part of the course again, I must be able to prevent those students from interfering with the success of others.
I cannot conclude this without mentioning the sheer joy that comes out of all this work. It is uplifting and humbling now to meet students on campus who took part in the course and express feedback. This coming year, with most of the course content now solidly covered with 15-minute videos, I will be in a much better position to follow the footsteps of my colleague Dr. Steffi Knorn and flip the course around. My vision is to be spending all of my class time helping students actually solve realistic problems in fluid dynamics, using modern tools and techniques. I look forward to 2021.