Muddling formative and summative assessment (#1326)

Topics/tags: Teaching

The other day, the CS faculty were chatting about various topics, and, at some point, the conversation turned to whether homework assignments should be formative or summative assessments [1]. For those not hep to the ed jargon, formative assessments are the kinds of work we give students to help them develop (form) knowledge, and summative assessments are the kinds of work we give students to assess whether they’ve learned the material. At least that’s how I understand them.

Depending on where you think learning happens, homework assignments might be formative or summative. If you believe the primary learning should happen from readings and classroom time, homework assignments could be summative. If you believe the primary learning should happen in doing the homework, homework should be formative, and you need to find another kind of summative assignment (say, an exam).

Of course, many of the courses in the CS department use a workshop-style approach, one in which most class periods involve students working together on problems. That kind of work is certainly formative, and it might encourage us to think about homework assignments as more summative.

Does it matter whether we think of homework as formative or summative? Definitely! How you think of homework affects the policies you associate with your assignments. If homework is summative, you want to ensure that it represents only (or primarily) the student’s work, and you forbid students to talk to each other about it [3]. If homework is formative, it seems reasonable to allow students to talk to each other. The latter means that there’s a risk that the work we see is only partly the student’s. However, we hope they learned along the way (and that they’ve cited the people they’ve gotten help from).

I treat homework as formative. While you start to learn from readings and lab, having to apply ideas in new situations (i.e., on homework) is where the material takes shape in your mind. And, as much as I’d hope that the readings and labs would provide sufficient guidance, I also know that many students benefit from talking through ideas with others.

I also believe [4] that we learn when we struggle with material. However, that struggle should be a comfortable struggle; students should feel they have the resources, including human resources, to complete the work. I’d rather encourage collaboration than discourage it.

That’s my perspective on the conversation. I respect that my colleagues may make different choices.

Where should the summative assessment happen, at least in CS courses? I prefer for that to happen in exams. These days, my exams take the form of sets of learning assessments (SoLAs); groups of small problems, each emphasizing one learning outcome. I suppose my exams have always had groups of small problems; in the past, they were less precisely targeted at individual learning outcomes [5].

Since exams and SoLAs are summative, students should not rely on other people (or generative AI) to answer them. We can partially enforce that perspective by having in-class exams and monitoring students during those exams. I’ll admit that I don’t like that model. In-class timed exams add stress; students may not demonstrate their understanding as well when stressed. So, I almost always give take-home exams with policies that limit what resources they can use.

Take-home exams provide additional opportunities for students to violate the policies and thereby misrepresent what they know [6]. A few students will take advantage of those opportunities. In my experience, it’s a small fraction. At least at Grinnell, I’ve found that students are here to learn. They’d like high grades, but they’d rather earn those high grades.

Whoops … I wasn’t planning to detour that far into my exam policies or reflections on academic integrity.

Let’s check the title of the musing. What was it? Oh, that’s right: Muddling formative and summative assessment. The approaches I’ve described above don’t seem to be muddled, do they? In-class lab work is formative. Take-home assignments are formative. Take-home exams and SoLAs are summative. That seems clear enough.

Here’s the thing. I use a mastery grading framework. So, if a student does poorly on a homework assignment or a learning assessment, they get feedback and the opportunity to resubmit. That turns each learning assessment into a formative activity, doesn’t it? That is, a failed learning assessment becomes a learning experience. Are SoLAs summative? When students are successful, SoLAs are summative. When students are unsuccessful, SoLAs are formative. It’s muddled. And that’s okay. I’m not sure there’s value in enforcing a strict distinction between formative and summative work.

Of course, I’ve never been good at enforcing such a distinction. Back when I gave more traditional take-home exams, I always tried to make exams a learning experience, one in which students had to apply the ideas they’d learned in new contexts. The best students always liked those problems. I even recall some telling me that my exams were their favorite learning experiences. However, that approach was not suitable for all students.

SoLAs may muddle (or muddy) the distinction between summative and formative assessment, but they do so appropriately. Perhaps all hard distinctions between summative and formative work are a bit of a fiction, as clear as driven snow.

[1] To be honest, I said something like, One of us [2] raised the question of whether homework assignments should be formative or summative. What do folks think?

[2] Not me.

[3] At least one of my colleagues who treats homework as summative forbids students from talking to anyone other than a faculty member about it.

[4] Perhaps even know.

[5] Back when I thought in terms of topics rather than learning outcomes, the problems targeted topics.

[6] I was going to write cheat. But let’s call it misrepresenting knowledge.

Version 1.0 released 2025-01-18.

Version 1.0.1 of 2025-01-20.