Hacking your final exam for alternative grading

The final exam can be a great platform for experimentation

Photo by Hans Reniers on Unsplash

Don't look now, but it's almost the end of the academic year. It's close enough to see summer break on the calendar and feel anticipatory joy. But it's far enough away to also see the last week of classes and final exam week in the same view. Which really kills the "summer freedom" vibe.

There is indeed an awful lot of work for us to complete before we can close things down until August. But in that work there are also some unique opportunities to take that work and make it more student- and growth-oriented. Especially for those who are using traditional grading methods but who are thinking about moving in an alt-grading direction, the waning weeks of the academic year are an outstanding time to experiment. And I think there's no better platform for experimentation than the Final Exam.

David posted some great ideas about final exams and the end of the term generally speaking a while back. I'd encourage you to check those out first. And if you are not giving a final, or if your final is not in your control (like a common final across a coordinated course) then take these suggestions accordingly. But if you are like me, and you give a final exam and you have some creative control over it, here are some ideas that can take your final from being the elephant in the classroom to an assessment that can drive your students' growth.

Use clearly defined standards

The first of the Four Pillars of alternative grading is clearly defined standards. This pillar states that growth-oriented grading methods should start by making it clear to students what they are expected to learn, by writing out content standards as actions that are clear and observable, result in evidence of learning, and which are bound to specific and important concepts. If you are using a final exam, it's presumably because you want to see what students have learned over the entire scope of the course. So there's no better time to get clear on those items.

First, write out for yourself what learning objectives you intend to assess on your final exam. Again, these should be clear to the student, observable (so you can actually evaluate their work), provide evidence of learning, and bound to important and specific items. In my current class on linear algebra and differential equations, for example, one of the objectives for the exam is "I can find the general solution for a first-order system of DE's with complex eigenvalues".

Being clear about your objectives for the final exam forces you to realize, among other things, that you cannot possibly include every single thing that students might have learned on it. You have to make choices on final exams and you can't do it all. Having the objectives in clear terms will help you make smart choices about the most important things to assess. For example, the objective I mentioned above involves solving a first-order system of differential equations, and ordinarily this requires students to compute the "eigenvalues" mentioned in the learning objective. Do I want students to do that computation on the final, or should I just give them the eigenvalues and let them build the solution from there? It's a decision I have to make which makes a big difference in the complexity (haha) of the problem, a decision I might not have even thought about until the standard was clear.

Even if you stop there, having clearly defined standards is helpful. But you can take it to the next level by publicizing the standards. For example, you can put all those standards into a document and give it to your students, and say "This [or a subset of it] is what you are going to be asked to do on the final exam". A growth-oriented approach to teaching means that we're not trying to trick or surprise students; we want to assess what they have learned, and it's only fair to make the target clear to them. A simple list of standards might be better for this than a "practice exam" where students have to reverse-engineer the standards from the exam items.

You might even consider putting the standards being assessed by each exam item, on the exam next to the item. For example on my current class' final, when I ask students to solve a first-order system with complex eigenvalues, I could preface that item with:

This item assesses the objective: I can find the general solution for a first-order system of DE's with complex eigenvalues.

That way it's crystal-clear to students why they are being asked this particular question. If you're concerned that putting the objective next to the item will give away the method for addressing the item, David mentioned that last week, and we say much more in our upcoming book.

Use a stealth EMRN rubric

Another pillar of alternative grading is the use of marks that indicate progress -- which doesn't typically include points for a lot of reasons. You might be stuck using (or still choose to use) points for different reasons. But it doesn't mean you can't sneak in a little alt-grading flavor.

The EMRN rubric is a four-level rubric that's come up in several of our articles in the past:

The EMRN rubric

While two-level ("pass/fail") and three-level ("Excellent/Satisfactory/Retry") rubrics are simpler to use, EMRN gives a little more nuance to the marks and involves a divide-and-conquer approach to assigning the mark, so it's still makes for fast marking.

Here's how you might use this rubric with a points-based final. Assume that the final exam is 100 points.

• Tell students that there will only be four grades given on the final: 50, 70, 80, and 100.

• Work that overall demonstrates thorough understanding of the concepts and which is complete and well communicated, will receive the full 100 points, even if there are a few relatively insignificant issues.

• Work that overall demonstrates thorough understanding of the concepts but which has a few issues or mistakes which are not relatively insignificant, gets 80 points.

• Work that is a good-faith effort to complete the entire exam and shows partial understanding of the concepts, but in which the work has several significant errors or gaps in reasoning, gets 70 points.

• Work that is missing, significantly incomplete, or otherwise does not show good-faith effort gets 50 points.

This is just the EMRN rubric, with the point values 100, 80, 70, 50 mapped onto E.M.R. and N respectively. You could include the text labels if you want, and there's no rule regarding the numbers -- if you want to use 100, 85, 75, and 60 instead then go ahead. Or, let an overall rating of “E” map to a “plus” added to whatever the course grade was before the exam, an “R” or “N” maps to a “minus”, and an “M” leaves the course grade unchanged. Or something like this. 

But the point (haha) is that you are not splitting hairs over point differences but using four broad "chunks" of points as an indicator of the overall quality of the exam work.

Doing the exam this way has several advantages. First, the grading should be a lot faster because you won't be stuck trying to distinguish between (say) a 72 and a 76, but rather just putting each exam into one of four buckets. Second, it reassures students that there is a "floor" in the sense that any student who shows up and puts in a good faith effort will earn no less than a 70 (or whatever you decide). Third, at the same time it makes the "ceiling" high: Students can make a few mistakes, but as long as the mistakes are minor and not too numerous, they will still get full credit.

Give the final exam twice

Conventional wisdom says that you can't do reattempts of a final exam because, well, it's the final exam and once it's done, the semester is over. A way around that is to give the exam twice: Once during the last week of classes, and then again at the ordinary final exam time.

Before you write this idea off, consider: If you use the "stealth EMRN" approach I described above for grading the final, in my experience you will find that the grading goes much faster -- so fast, in fact, that it's reasonable to think that in many classes, you could get final exam version 1 graded and returned by the start of final exam week if you gave it at the beginning of the last week of classes. You can even use the opportunity to give helpful feedback which is another of the Four Pillars we haven't discussed (because feedback isn't really helpful if there's no possibility of reattempt).

Here's how this can work:

• Make out a final exam "version 1" that has clearly defined standards and which is minimal. That last bit is crucial: This doesn't work if you don't keep things simple and minimal on your exam, and simplicity/minimalism is a good idea for exams even if you do nothing else in this article.

• Give version 1 at the beginning of the last week of classes. Because let's face it: Nothing else productive happens during this week with the possible exception of project presentations. (If you have new material planned for the last week of classes, quit kidding yourself and just erase that from the calendar.) You might even consider making version 1 a take-home exam.

• Grade it using the stealth-EMRN "chunked points" approach and get it back to students by the end of the week.

• Then make up a "version 2" of the final, with the same standards (or drawn from the same superset of standards), to be taken in person during the final exam period. (Or take-home, if you like.) Students who got 100/"E" on the first attempt don't have to take version 2.

• Grade it the same way and let students keep the higher of the two scores.

There are a lot of questions about details here: What if a student got a problem completely right on version 1, if they only got 80/"M" on version 1 then do they have to redo everything? And so on. I'm afraid I am not going to address all those issues -- or any of them, actually, but rather just say: If this sparks some interest, work the details out in a way that makes sense for you.

---

I've written before on my main blog about ways that I am attempting to update what my final exam actually does: Substituting particular mathematical problems for creating mind maps of the semester and writing reflective essays, for example; or even just not really giving a final exam at all. But even if you are giving a final exam -- as I am doing for my particular class this semester -- it doesn't have to be such a stressor for students or a drag for us. Use it as an opportunity to dip your toe in the alt-grading ocean and orient the whole thing toward growth.