Today’s post is a response to a question I got on Google+ this week, about cumulative exams in specs grading:


The first time I used specs grading was one year ago. I thought it worked very well despite being the first time I had ever used it. The one thing that I felt was a major weakness was that there was no way to check whether student attainment of course objectives persisted past the point of assessment. That is, I was never sure that when students demonstrated they had mastered learning objectives that they had really learned something for the long term, versus just studying for a quiz and then forgetting it.

In last year’s classes, we had split the learning objectives of the course into about three dozen distinct measurable outcomes like Determine whether a relation is reflexive, symmetric, antisymmetric, or transitive and each learning objective had its own mini-assessment. We set aside a few days in the course for assessment periods and students would come in and take whatever assessments they wanted or needed for that day.

What I noticed was the following: A student would come in on an assessment day and take an assessment over a learning objective – let’s say objective M-4, and Pass the assessment. But then, they would fail to write down that they had passed M-4 already; or they would have that information written down but would forget to make a note for themselves on the next assessment day. Either way, they would mistakenly take the assessment for M-4 again – and fail it!

That’s not good, and this happened way more often than I felt comfortable with. So it made me think that I need some kind of big-test-at-the-end to ensure that what they learned earlier had “stuck”. Normally this is taken care of by a final exam. But I don’t like final exams: High-stakes, one-and-done timed tests are not an authentic form of assessment, nor are they particularly humane or even just plain useful. As Zdenek says, they go against the philosophy of try → fail → try again → pass.

So, here’s what I instituted this semester in an attempt to rectify this. It might be helpful to follow along with the syllabus to track all this.

The specs grading system I have in place now has a number of components and one of them is timed assessments. I have given all the details in this post. Briefly, each course I teach covers five main topic areas. Each of those topic areas has two timed assessments that go with it: A Level 1 assessment over basic skills and a Level 2 assessment over more advanced concepts. They are graded on a two-level rubric, Pass and Repeat. A student who earns a Pass on an assessment is done with that assessment. A student who earns a Repeat can take a new version of it at ay future assessment period (we have set up six of these throughout the semester). The grade for the course that the student earns is partially tied to the number of assessments passed at each level.

So this is sort of like last year’s system, except instead of having each learning objective with its own mini-assessment I am lumping together objectives basically into “easy” and “hard”. I’m liking the way this has worked out – much less work for me and for the students, much less overhead and fewer logistical issues.

As for the summative end-of-semester experience: I have set aside the entire last week of classes for what we call recertification exams. Each main topic area will have a recertification exam that is a truncated version of the original assessments. Students have to recertify on a topic in order to progress toward a particular course grade. For example, to earn a B in the class, students have to Pass the Level 1 assessment on all five topics and the Level 2 assessment on any two topics of their choice; and pass a Level 1 recertification on all five topics and a Level 2 recertification on the two topics they chose earlier. (There are other things to do as well. See the syllabus.)

Like I said, we have scheduled the entire last week of classes for taking recertifications. The final exam period is also for taking recertifications. That’s enough time (almost 4.5 hours) to reasonably get through seven of these recertifications – remember they will be truncated from the originals – and have time to study in case of a Repeat and there will be up to three chances to repeat if necessary. And there is nothing else in the course going on during that time – there’s no final exam or final project. Just recertification.

I don’t know how this is going to work out in real life. But I think it’s a good response to the final exam situation. It prevents students from dropping all their knowledge as soon as the assessment is over, but it carries the try/try again ethos with it as well.

Credit where due: Justin Dunmyre came up with this recertification idea and when he explained it to me at a conference this summer I knew this was the model I wanted to try.