Six things I no longer do with alternative grading

An exploration of addition by subtraction.

Photo by Valerie Sidorova on Unsplash

Higher education is very good at addition, but struggles with subtraction. The default of higher education and many of those who work within it is to add stuff on top of other stuff, whether or not the cost of addition outweighs the benefits, and to continue this semester after semester. We know how to reduce and how to say “no”, but it’s not what we were trained for and it’s not in the nature of the system.

But the older I get and the longer I spend as a professor, the more focus I have on radically simplifying everything that is around me. Over the last few years as I continue to learn how to grade alternatively, I have been specifically focusing on what I can take out of my practice rather than what new, shiny things I might put in.

Today I wanted to share six things that I have removed from my teaching practices that are related to alternative grading and explain the consequences – good, bad, and ambiguous – of doing so.

Having too many learning standards

I’ve mentioned in some earlier posts that my first attempt at specifications grading involved having over 60 different learning standards in my course. Those 60 standards covered every possible topic that I felt students could or should learn in the course. And also, obviously (to me, today) created an enormous and unending amount of grading for me and work for my students. I knew by around the third week of the semester that 60 standards was way, way too many.

So in the following semester, I was much pickier about my standards — not about the rigor of the standards or the thresholds for acceptable work on them, but about the standards themselves. I started saying no to a lot of topics in my class that I formerly felt were inviolable and must be assessed in order for the course to be “a real course” in the subject. Either those topics would get cut out of the class entirely, or they’d be included as interesting side quests but not to be assessed. For some topics that were so small they honestly didn’t merit their own assessment, I would either not assess them at all or glue them together with other similar standards to form one mega-standard that would be assessed¹.

Over time, the number of learning standards in my courses has dropped from a completely insane 60 down into the 20s and eventually to 15 or 16. This is a nice round number because it’s the number of weeks in a typical US college semester. And if you have 15 learning standards, that works out to one per week on average. You can use this structure and this number of learning standards as a time-saving device when you are planning out the lessons on your course calendar: Simply target one week per standard.

This semester I’ve managed to cut my standards down to 12. I think 12 is even better than 15 because then you can line up your standards with a 12 week semester which I will write about below.

Cutting and consolidating this way does mean you are covering or assessing fewer things. This can cause problems for some faculty, so proceed with caution. The Chesterton’s Fence principle applies here: Don’t remove things until you understand why they were there in the first place.

Using multiple layers of assessment to certify mastery

As if it weren’t bad enough that I had 60 learning standards in that first specs-graded course, each standard required three successful quiz attempts — three “layers” of assessment — in order for me to certify if the student had actually learned the standard. This seemed like a genius idea… at the time. But several years ago I ended this practice, and now have a higher bar for “acceptable” work and a rule that a single “acceptable” attempt on a standard was enough.

There is value in requiring more than one successful assessment on a standard in order to certify the student has met the standard. It’s possible for a student simply to have a good day, or to get lucky, and demonstrate mastery once, but having them do it twice or possibly three times seemingly allows you to triangulate the position of the student with greater fidelity. I know that in the past when I’ve used multiple layers of assessment in this way, I’ve seen students master a topic in week 3 and then fail to demonstrate mastery in week 12, so the second layer prevented (or at least delayed) a false positive.

However, it’s undeniable that if you require n levels of assessment, it is going to multiply your workload roughly by a factor of n, and not just your workload but also students’. I have to believe that having an inflated workload cuts into the helpfulness of feedback. It’s just harder to give helpful feedback when you’re exhausted. And anyway, are we really sure that multiple levels of assessment improve the validity of the data that we get? It feels like this should be the case, but I have never seen an actual statistical analysis to prove that it does. Maybe, in fact, one layer of assessment is just as good as three. Even if one layer were only 80% as good as three, I would be willing to risk it and just do one, because I am going to be a better grader and professor if I am not being hounded by my own grading workload.

Flexible due dates on assignments

The conventional wisdom is that having flexible due dates is a sign that one truly cares about students. But having been publicly supportive of flexible deadlines in the past, even advocating the outright elimination of deadlines, I’ve changed my mind about this. (And I think I still care about students.)

The last time I had flexible deadlines was three years ago. If a student found it difficult to meet a deadline, they could just fill out a form that notified me that they were picking a different deadline, what that deadline was, and give their reasoning and anything else they needed for support. Sounds good, right?

Unfortunately this problem caused more problems than it solved. Some students used it merely to deal with procrastination.² On the other hand, some had legitimate needs for flexibility but wouldn’t use the policy, for a variety of reasons.³

My students were asking for deadlines by the end of the semester. I think they were right. I found that, while students value having flexible deadlines, they worked better and learned better when there were fixed deadlines that were set reasonably and enforced fairly (along with reattempts for work that was submitted by the deadline but didn’t meet the standards). They are a stake in the ground on their calendars that gives them structure, and for my students at least, the structure was better than so much flexibility.⁴

Graded asynchronous work

As I’ve written extensively here, I’ve recently shifted to a model in my teaching where anything that is done outside of class is graded on completeness and effort only, getting only feedback and not a mark. This is basically ungrading on a small scale and is intended to get students to build their mastery of a concept through iteration with a feedback loop. But the “final” demonstration mastery is done in class through timed tests. These often replicate or are significantly based on the asynchronous work that students do, but it’s the tests that are graded, not the asynchronous work.⁵ (The “final” is in quotes because there are limited reattempts built in; see my syllabus for details.)

I have been and remain fairly critical of the concept of ungrading as a whole, but I readily admit that this shift has really been beneficial for me and my students. I originally did it to mitigate issues with generative AI, and I think it’s been successful in that. But it also motivates a nice shift towards a culture of self-regulated learning in my classes that I also like very much.

This approach has several upsides so far. First of all, to be perfectly honest, I have much less grading to do because fewer students are going to turn things in if there’s not a “real grade” on it. I don’t like this, but I won’t lie: I do like the lighter workload. Less facetiously, this kind of grading is more the sort of work that I want to do with my students. I am getting good faith, honest attempts from students who are motivated to learn from feedback loops, and I am giving those students helpful feedback and seeing the fruits of iteration with it.

The main downside here, as you might guess, is that a lot of students skip doing the ungraded work and just try their luck in the timed test. This often works out roughly how you would expect. There are re-attempts built into these tests, but they’re very limited because they take up time and space. So students almost by default tend to put themselves in bad situations and the students who are in the worst situations tend to be the ones who need reattempts the most.

Covering any content at all in the first week and last two weeks

One of my most popular posts from my old blog is called The 12 Week Plan. You can read the whole thing at the link, but here’s the gist: You design a 15-week course to fit entirely into 12 weeks, then use the extra time for initial onboarding in week 1, and two solid weeks of reassessment at the end. I have been a devotee of the 12-week plan for years now and saying “no” to Covering All The Things, especially in weeks 1 and 14-15, has been one of the best moves of my career for student success.

The upside of the 12-week plan is that you get these three weeks of time to do useful stuff. Students can learn the grading system in the first week of class and enter into the main body of the course confidently, and more importantly they are able to put your grading system into a background process and not obsess about it. And there is far less of an insane crush at the end of the semester to try to assess and reassess on topics, especially those that first appear late in the term. And let’s face it, by the end of the semester, you’re working with students who are in no shape to absorb new material. So you might as well just use that time profitably by giving students time and space to breathe and reattempt what they need.

The downside is that you lose three weeks of time to do course material. I don’t necessarily see this as a downside. But it can be hard, in some cases impossible to implement fully. One engineering course from a couple of years ago was so packed with topics that were deemed “essential” for subsequent courses that all I could manage was a “13.5-week rule”. But we work with what we’ve got, and even keeping just the last week blank on the calendar for reassessments might be a huge help for students.

Worrying about downstream academic performance

I get asked the following question a lot: Do students who have alternatively graded courses perform better or worse or the same as their counterparts in a traditionally graded course, in later courses? For example, how do students who have a specs graded Calculus 1 course do, grade-wise, in Calculus 2 compared to students who had a traditionally graded Calculus 1 course?

This isn’t a bad question, but I have never felt motivated to research the answer seriously because if I’m being perfectly honest, I just don’t care that much about it.

Let me explain: I care a lot about student success. Specifically, I care about the success of the students that I currently have in front of me nine hours a week, in the classes I am teaching this semester. As for the students I don’t yet have, or the grades of my current students in future courses, I somewhat care. But I also must admit I have no control over this. And I have a strict policy of not getting invested in things over which I have no control.

So instead, I’ve learned to trust myself and my ability to design courses, and trust my students to be able to learn well in the future once they have been given the tools. It is possible that because of the reductions and subtraction that I’ve mentioned above, that my students are missing some topics or haven’t been assessed as thoroughly on certain topics, as they could have, and that some of those topics might be the most important things in the world to the next professor up the chain. But the opposite is equally possible. Therefore, it’s just not worth the energy to think about it. That energy is better spent invested in the students that I have now, and the courses that I am teaching this semester, and remember that “sufficient for each day are the rigors thereof”.

Addition by subtraction

I’m a firm believer in the concept of addition by subtraction. We need to be careful not just to add things into our systems simply because they might provide some small value. The value that’s provided by the things we add, has to be balanced against the values of the things that we can no longer do because we’re adding.

Can every person reading this article do every single thing that I have mentioned here? Probably not, nor should they necessarily. But I do think it’s universally true that a simple system with a small number of moving parts is likely to be a better fit for you and for students than a complex machine that seems to cover it all and produce seemingly perfectly optimized results.

1. I wrote some more about this in my recent post on writing clearly defined standards.

2. I asked them, and this is what they told me.

3. For example, some felt it was a point of personal honor to complete work by the deadline, no matter what; others felt it was inappropriate to ask even though I clearly wanted them to. Student beliefs about deadlines are hard to change.

4. I also stopped using tokens, even though I am using a variant of specifications grading and tokens are considered a core feature of this approach. Because of other changes to my course design, there was just no “market” for tokens anymore, so they became a deprecated feature.

5. The asynchronous work earns engagement credits.