Assessment in education: what it's actually for
Ask a room of faculty what assessment is for and you'll get two answers. The official one: measuring whether students met the learning outcomes. And the honest one: producing a number by week twelve, because the system needs a number.
Both are true. The gap between them is where most course design quietly goes wrong.
Assessment in education is simply the evidence you collect that learning happened. Not the grade — the grade is what you do with the evidence afterwards. Once you separate those two things, a lot of familiar problems start to look different.
The question that sorts everything out
Before format, before weighting, before rubrics, one question: what will this piece of work let me see that I couldn't see otherwise?
If the answer is "whether they did the reading," you want something small and frequent. If it's "whether they can act on this when the situation is messy," you want something that gets messy. And if you can't answer it at all, you've found an assessment that exists because it was there last year.
That last case is more common than anyone admits. Courses accumulate assessments the way desks accumulate paper.
The three jobs assessment does
Every assessment does at least one of these. The good ones know which.
It tells you what to do next. Low-stakes, quick, often ungraded. You find out on Tuesday that half the room has misread the core concept, and you fix it on Thursday — rather than discovering it in January when nothing can be done.
It certifies that they can do the thing. Higher-stakes, later, graded properly. This is the evidence that goes on a transcript and that an external examiner might want to look at.
It makes them learn. Underrated, and the one people forget. Students study what gets assessed. The format of your assessment is a set of instructions for how to spend the term — which means a badly chosen one actively teaches the wrong habits.
The split between the first two is what people mean by formative and summative, and it's worth getting straight: here's what each type is for and when it earns its place.
Two ways an assessment fails
An assessment can be beautifully written and still be broken. There are only two ways.
It measures the wrong thing. Your outcome says students will evaluate competing approaches. Your assessment asks them to describe three of them. Everyone passes, the numbers look healthy, and you have no evidence that anyone evaluated anything all term.
This mismatch is almost always in the verb. Put your outcomes in one column and what each assessment actually asks students to do in the other, side by side. Half the gaps show up immediately — and usually you'll find one outcome with no evidence at all and another with four assessments piled on top of it. Designing the task backwards from the outcome is the fix, and it's less work than it sounds.
It measures inconsistently. The same work would get a different mark on a different evening. Not because you're careless, but because holding a standard steady across sixty submissions from memory is not a thing humans do. That's what a rubric is for — and it's also why the last paper of the pile gets marked differently from the first.
Validity and reliability, if you want the formal words. Measuring the right thing, and measuring it the same way twice.
Short on time?
Skip the rest of this guide — let Quindaria build the rubric for you instead.
We'll email you a confirmation link. Click it and you're in — it takes a second.
What AI changed, and what it didn't
It didn't change what assessment is for. It changed which formats still produce evidence.
If the task can be completed from public information alone, it now tells you very little about the student in front of you. That's not a moral problem to be policed — it's a design problem to be solved. The assessments still doing their job have something the model doesn't have: your data, your cohort's own work, a context specific enough that generic output is visibly wrong, or a requirement to defend the choices out loud afterwards.
The point isn't to make work AI-proof. It's to make sure that whatever you collect is still evidence of their learning. The same logic applies to using the tools yourself: there's a way to use AI in course design without lowering the standard, and it comes down to what you feed it.
The over-assessment trap
More assessment is not more rigour. Every extra graded task costs you marking time and costs students attention they could spend on the work that matters.
Three tests for anything currently on your course:
- Would you notice if the grade weren't there?
- Does it tell you something the other assessments don't?
- Would you defend it to a student who asked why it exists?
Anything that fails all three is administrative habit, not assessment. Cutting it usually improves the course and always improves your term.
Where to start
You don't need to redesign anything this week. Take one module and write two columns: outcomes on the left, what each assessment actually asks students to do on the right.
You'll see it straight away — the outcome nothing measures, the one measured four times over, and the task whose real subject is formatting. That afternoon is the highest-return course design work there is, and it doesn't require a committee.
Then build the tasks that close the gaps. That's the part Quindaria was made for: give it your course outcomes and it builds activities and rubrics that assess what you actually claim to teach, in the format you choose. See a real activity →
Your course. Your style. Always aligned.
Pick one module. Two columns. You'll know by the end of the afternoon what your assessment is really measuring.
See a real grading rubric
A complete rubric built from a real assessment — every criterion and every level written out, ready to use in Word or Excel.
We'll email you a confirmation link. Click it and you're in — it takes a second.
We'll send it straight away. Nothing else, and you can leave any time.
Any email works — use the one you actually read. What we do with it
Keep reading
Types of assessment in education, and when each one earns its place
Diagnostic, formative, summative, ipsative — the vocabulary is familiar, but the useful question isn't what each type is called. It's what each one is for, and what it costs you to run it.
4 min read
Grading rubrics: the complete guide
What a grading rubric actually is, why a vague one costs you more than it saves, and how to build one that holds up — for any assessment, from a first-year quiz to a final-year dissertation.
6 min read