I’ve written before about my experience teaching my graduate course this semester, but I haven’t talked about one big experiment I tried: working towards standards-based grading. I started hearing about different implementations of this over the last year or so, and thanks to a Project NExT panel at the joint meeting from Eric Sullivan and Benjamin Braun (links go to their slides), I thought I’d take the plunge.
The idea behind typical standards, specs, or mastery-based grading is to track students’ progress with mastering skills directly, instead of filtered through a weighted average of percentages on assignments. In addition, students have multiple opportunities to improve their work and demonstrate mastery of these skills. Final grades are usually assigned based on how many skills have been mastered.
I didn’t do a full implementation of standards-based grading this semester, but I did get a good start. I skipped the really hard part of this – listing all the learning objectives for my students – because I’d never taught the course before and some of my goals for the class felt kind-of nebulous. This class is algebra for teachers, so a lot of the course is about teaching students to think abstractly, generalize, and write simple proofs. Writing these goals in simple, assessable terms seemed like too much work for a first attempt.
Instead, I started with the easy part of standards-based grading: the actual grading part. I graded each homework problem on a scale of 0-2. A 2 is mastery. Maybe not completely perfect, but it’s clear that the student has demonstrated that they have the skills I’m assessing. A 0 is work that is completely on the wrong track, and 1 is somewhere in between – maybe undeveloped or incomplete, but with a good start. I liked this scale a lot, because it reduces the variance in my grading: if I try to grade a problem out of ten points, the difference between a 5 and a 7 might depend more on my mood than on the actual quality of the work. But the lines between a 0, a 1, and a 2 are almost always clear.
I had a very small class, so I let my students resubmit their work as many times as they wanted. Others put a time limit on resubmissions to prevent a glut of grading at the end of the semester. But the idea is that students can keep refining their work until they are proficient at the skills we want them to learn. And it forces students to learn from their mistakes instead of just shoving a bad grade in the back of a folder and never thinking about it again.
I assigned final grades based on how many 0s, 1s, and 2s students had by the end of the semester. An A was more than 95% 2s, a B more than 85% 2s, C more than 75%. Students knew exactly what they needed to do in order to get the grade they wanted and (most) would resubmit accordingly. One student was terrified that she was going to fail in the beginning of the semester, but after working with her extensively she ended up bringing her grade up to an A.
I have to admit, this was not an unqualified success though. These were graduate students, most of them current teachers with families, so I was very lenient on deadlines. I figured everybody would turn in the required resubmissions by the end of the semester without a lot of hand-holding. That was not the case. If I had this to do over again, I would have been more proactive and made sure that every student knew if they were underperforming.
I will definitely be doing this again for my smaller classes. The simpler scale saved a ton of time and effort, so the re-grading didn’t feel like that much extra work. Next time I’ll even try to align my assignments with learning objectives. I’m also interested to see how undergraduates respond to this method – my guess is that they’ll jump on board a little more easily, but I’m not sure. I’ll report back next year.