taking specs seriously

I’ve been an advocate of standards-based grading since I started using it over three years ago. It has addressed many of the concerns I had about the dominant point-based grading system and encouraged students to move forward in their understanding rather than feeling trapped by past performance.

I’m not solely an SBG proponent when it comes to grading, however. For one thing, I find it hard to adapt SBG to upper-level math courses. For another, the time seems ripe for experimentation in grading practices as more of us realize the shortcomings of what we have inherited from decades past. Not that we should constantly reinvent the grading process, but we should be open to various thoughtful ways of providing authentic assessment.

So I was certainly interested a couple of years ago when several fellow instructors began talking about specifications grading, a method espoused by Linda Nilson in her book Specifications Grading: Restoring Rigor, Motivating Students, and Saving Faculty Time. I adopted some of the ideas I heard and appreciated the increased flexibility it offered.

However, it was not until this summer that I read through Nilson’s book. It was useful because it seems Nilson and I think differently in ways I can’t quite put my finger on, and so the book has lots of ideas I would not have intuited on my own. Here are a few of the things I garnered from reading the book that I hadn’t picked up from online discussions (not that these things weren’t said, but this time they stuck):

• Sometimes it’s OK to use percentages. I’ve been highly points- and percentages-averse since starting SBG. Percentages, my argument went, were essentially meaningless, because they’re constantly being curved (so they don’t really represent a “percentage” of anything) and the difference between 80% and 81% is essentially a coin toss (so they aren’t as linearly ordered as people like to think). But that argument isn’t uniformly true. In a course where precision is important, it is possible to measure, for instance, how many definitions a student can correctly state. For my upcoming analysis class, I expect “A” students to get definitions right 95% of the time, “B” students 85% of the time, “C” students 75% of the time. This really is quantifiable, and a definition is either correct (with respect to the established conventions of the subject) or not, so each one can be graded yes/no. As long as not everything is forced into a percentages model, this can be an effective way to give feedback.
• Make students work for an A, but give them some choice in how to get there. As instructors, we want an A to represent mastery, an indication that the student can think nimbly and complexly about the subject. Ideally, students who earn an A will be the ones most invested in the subject. To demonstrate all this, students should have ownership of their work. They should make meaningful choices that reflect their interests and their skills as well as the subject at hand.
• Not everyone has to do everything. This is closely tied to the previous point. Nilson uses the metaphor of “hurdles”: grade levels can be differentiated by having students clear either higher hurdles (more complex, better quality work) or more hurdles (more extensive work), or a mix of the two. I’m not generally a fan of having students earn higher grades by just proving they can do more—that takes more of my time, and more of theirs. But true mastery requires a measure of initiative. Having a small number of optional assignments that give students opportunities to distinguish themselves makes sense as part of a larger grading scheme.
• There are good reasons to limit reassessments. Of course, one of these reasons is the subtitular “saving faculty time.” In past upper-level classes where I’ve allowed essentially unlimited resubmission, I’ve been swamped/behind at several points in the semester as students frantically tried to get something accepted. But that’s not even the best reason. By limiting reassessments and grading work pass/fail (or pass/progressing/fail or some other variant), students are encouraged to submit their best work each time, and to spend extra time making sure they check its quality before asking me to do so. The onus is on me to establish clear expectations, and on students to meet them. We’re not negotiating what’s acceptable through repeated revision and grading.

I also found the chapter on cognitive models (Chapter 3, “Linking Grades to Outcomes”) helpful in considering what it means to have a higher level of mastery; previously I wasn’t really familiar with anything beyond Bloom’s Taxonomy.

If this post was of interest to you, I hope you’ll consider joining the Google+ Community on Standards-Based and Specifications Grading” (SBSG) Slack workspace on mastery grading, where teachers of diverse disciplines are meeting to discuss how to implement these two particular alternative forms of grading.

Tomorrow I’ll share my full set of specifications for real analysis.