Simplicity and complexity in grading
Traditional grading — the form of grading where student work is submitted, evaluated, that evaluation converted into a numerical score, and then the numerical scores eventually statistically combined to produce a letter grade for the course — has lots of problems. It seems like most of those problems stem from oversimplification.
Because traditional grading assumes that learning can be reduced to something measurable on a numerical scale, we end up with data that convey no information. Say a student makes a B in calculus. What does that student know about calculus? It’s impossible to say, because there are collisions — students whose mastery and knowledge of calculus are vastly different but who end up hashing to the same grade.
It should be no surprise that the primary conversation that grades drive is not about the content of the course or the nature of professional standards of quality. The primary conversation that traditional grades drive, is about grades themselves. Example: Today I worked at our downtown campus, surrounded by students all day long outside my teaching role — on the bus, in the Starbuck’s line, at the library. Most of them, if they are talking about school at all, are not talking about what they know or what they need to work on: They are talking about points and letter grades. If it weren’t for bad grading systems, they might not talk about school at all.
And it’s all because of this oversimplification of the learning process and student work.
And yet: We’re used to simplicity when it comes to grading. When we open a syllabus, we expect to see the grading system encapsulated in a single table that shows the items for assessment, how many there are, how much they count toward the grade, and how the letter grade is put together. All in one small portion of the page. If we try to institute a grading system that is more complex than that — if, for example and God forbid it should occupy an entire page of the syllabus — then the system will be branded as “too complicated” and many will never see past that.
This is the danger that specifications and standards-based grading encounters. And for me, it’s been the hardest thing about using specs grading in my classes so far, a year after I started.
With standards-based or specs grading (which I’ll put together and abbreviate as SBSG) we are saying: Learning is more complicated than a numerical scale. Here is a system that takes much of that complexity into account and will give you a much more accurate and usable grade. Which is true! But we pay a price for the realism that SBSG provides, in the form of greater complexity.
This semester I used specs grading again in my discrete structures courses. I worked very hard on this “third iteration” version of specs grading to make it the best version yet. I figured out a system of certifications (adapted from Justin Dunmyre, but none of the following is his fault) that would provide solid evidence of student mastery, all using the classic specs grading tenets of rich feedback, generous revisions, and adherence to professional standards.
And it went over like a lead balloon because it was too complicated.
I knew it was too complicated when I wrote the syllabus. It occupied six pages of my syllabus and even then, I wasn’t satisfied that I had explained it well enough so I made a video to accompany the syllabus, just about the grading system, that ended up being 23 minutes long. It boggles my mind that I could look at the syllabus and that video and not think that I was going to have issues with this.
Students, for their part, have played along gamely with my system. But the system is so bloated that student conversations about their assessments have been mostly — you guessed it — about grades, even though we don’t really use points for anything. It has ended up being just a wierder version of traditional grading.
So, all of us who use SBSG have to balance the following concerns:
- When a grading system is too simple, it’s easy to use but it provides corrupted or unusable information about student work, and causes students to focus on grades rather than on learning.
- When a grading system is too complicated, it provides a high degree or realism about student work but the complexity drives off students and makes it hard for them to focus on learning.
What I’m finding is that SBSG has a strong tendency to fall into the second category. Maybe it’s because people like me who like to come up with new “systems” in our classes enjoy systems so much that the systems we make are elaborate, sometimes like Rube Goldberg machines. It’s hard to commit to finding the balance between simplicity and complexity but that’s where the sweet spot is for student learning with our grading system.
Originally published at rtalbert.org.