The Chronicle of Higher Education had a great post this past week about Specifications Grading, a style of grading that moves away from holistic (“too light–feels like a C”) approaches into the world of contract grading, where specific criteria are set that, if met, earn students a particular grade.
I hate grading, so I was eager to read more. This style of grading sets clear, achievable metrics for success on a given assignment, and ties that success to the learning outcomes for both the assignment and the course. Grading itself becomes simple–either students met the metrics, or they did not. Students are given more choice in how to approach their work for the course, but at the same time, must complete enough work at a satisfactory level in order to pass the class or get their desired grade. This system, if done correctly, can eliminate student grade complaints, put students in charge of their own learning, AND reduce faculty grading time (YES!), What’s not to love?
Specifications grading, as outlined by Linda Nilson in her 2014 book, Specifications Grading: Restoring Rigor, Motivating Students, and Saving Faculty Time, takes contract grading further, setting clear criteria for ‘satisfactory’ level work on individual assignments, with groups of assignments bundled together around particular grades. There are three major components to this process:
- Clear assignment specifications, tied to learning outcomes
- Bundles of assignments tied to grades
- Tokens for ‘redemption and flexibility’
First, every assignment in the class is given clear specifications regarding what will be considered satisficatory-level completion. These will be parameters appropriate to the assignment–for an essay, that might include word count, absence of spelling/grammatical errors, a certain number of scholarly sources, appropriate citations, presence of a clear thesis, and a clear connection between thesis, argument, and evidence. Any student who achieves these (or whatever criteria you set) receives a ‘satisfactory’ on the assignment, while failing to meet any single criteria is ‘unsatisfactory’. Per Nilson:
In sum, complete, satisfactory work receives full credit (full value), and incomplete, unsatisfactory receives no credit/value. For students, it’s all or nothing. No skipping the directions and no sliding by on partial credit for sloppy, last-minute work.
This creates a higher bar for a satisfactory rating, but one that is more easily graded than the ‘feels like a B+’ approach, or even a standard point-based rubric. It also requires us to be very honest with both ourselves and our students on what is considered passing level work, and how an individual assignment is tied to the learning outcomes for the course. If done correctly, however, this approach should minimize the number of questions about the assignment and student anxiety about completing it.
These assignments are then bundled together, with grades tied to achievement of particular bundles. A bundle consists of one or more assignments completed at a satisfactory level. They can be created according to learning outcome or the challenge level of the work. To get a particular grade in the class, students must meet all the criteria of that bundle. Students therefore can choose which bundle they wish to achieve, putting the decision entirely in their hands about what grade they want to aim for, and giving them extremely objective, measurable criteria on how to achieve that grade.
In her Inside Higher Ed article, Nilson describes a few ways to bundle. In one case, you take ten assignments or tests, and then rate each one–perhaps whether it is easy or hard, or perhaps based on Bloom’s taxonomy of learning. In her example, satisfactory completion of bundles 1-5 would earn a D, 1-6 a C, 1-7 a B, and 1-10 an A. In a simpler system, assignments are grouped into four bundles, rated from basic to highly challenging, with each bundle tied to a letter grade. Jason Mittell’s system has a C level bundle that includes attending courses with no more than 5 absences, completing 7 weekly response papers at the satisfactory level, and completing all take home essay questions (either advanced or basic question option) at the satisfactory level. Higher bundles reduce the number of allowed absences, increase the number of weekly response papers, and require attempts on the advanced questions on the take-home exams, with the A bundle also requiring an additional assignment.
The bundling system gives students flexibility in how they approach the class, but still ensures that a student must complete the course learning outcomes in order to pass. As Nilson points out, it requires us to clearly tie our grades to student learning:
Let’s admit that, right now, our grades have little connection to outcomes. Students earning an A may have achieved all the outcomes of a course, but what about those getting a B, a C or a D? Did they achieve some outcomes and not others? If so, which ones? Or did they achieve few or none at an acceptable level? Even so, they passed the course.
The final component of this system is the creation of a token economy. Students are given a set number of virtual ‘tokens’ at the start of the class and can turn them in for various passes throughout the semester determined by the instructor. Options might include: turning an assignment in late, skipping an assignment entirely but counting it towards a bundle, eliminating an absence or lateness from the record, or revising an Unsatisfactory assignment. Students can also be given an opportunity to earn additional tokens by attending talks and events, turning in rough drafts of work for feedback, participating in class, or any other behaviors you want to incentivize. This token economy can help reduce requests for additional time and forgiveness for late work, as students can just turn in their tokens with no comment or stress.
The three main downsides to the Specifications Grading system are the initial investment of time on the part of the instructor, the lack of differentiation between satisfactory and exceptional work, and the potential for limited feedback. On the first point, I am a big believer in putting in the work on the front end to save time on the back end. Yes, changing my assignments, linking them very closely to the course learning outcomes, writing out the specifications, and creating bundles will take time. But it will make grading itself much quicker once the term starts, and this investment is only necessary once–the grading time save will remain in additional sections. Coupled with the benefits to students, I do not find the time issue a compelling objection.
The second issue is motivating excellence. If we want our students to excel, ultimately, then are we incentivizing mediocrity in this system by focusing only on satisfactory work? I think that smart design can handle this problem. One method might be to include an assignment that requires higher-order thinking and exceptional work just to be considered ‘satisfactory’ and reserve ‘A’ grades for those that complete that assignment, with an A- the highest grade for those that do not do this. Another might be to award additional tokens for exceptional performance on a given assignment. A third comes from Robert Talbot of Grand Valley State University, who uses an EMRF rubric that gives room for more nuance than just pass/fail or unsatisfactory or unsatisfactory. To get an A in his class, students must earn an ‘E’ (exceeds expectations, or excellent work) on approximately 25% of their passing assignments.
Finally, this system can result in limited feedback for students. Grading holistically or with a simple rubric might compel an instructor to give more feedback in order to justify their grade, so a pass/fail model like this, based on specifications, might limit that feedback. I think this can be a danger, but depends on the execution of the instructor. For students that get an unsatisfactory rating, simply indicating which criteria were not met is itself feedback, and fairly easily given–you can circle the neglected criteria on the list of specifications and attach it the assignment. Another option is to build in developmental feedback into the course, such as a mid-semester report that notes the strengths and weaknesses they’ve demonstrated so far. Finally, nothing stops instructors from giving extensive feedback from this system–feedback itself can be divorced from grading, and still given in droves according to instructor preference.
Overall, I found the articles on this subject very compelling and I plan to rework one of my fall classes to try this method out. I will report back here to ALPS on how it goes. In the meantime, I recommend you check out the articles I linked in the text above, particularly Nilson’s piece in Inside Higher Ed, and Mittell’s extensive piece detailing his assignment specifications and bundles, as well as his update on how the students are responding to the new system. Also check out Robert Talbet here, an update here and here . He talks extensively about the key role of failure and trying again as part of the learning process, something we talk up quite a bit here on ALPS. He also manages a community on Google Plus on specs grading.