Formative assessment has a marketing problem. The name sounds like a policy document, and most training on it describes something that takes a lesson to set up. In practice the useful version is much smaller: a two-minute check that tells you whether to move on or go back, answered before the information stops being actionable.
The constraint that kills most of these ideas is marking. A check that generates thirty scripts you have to read is not a five-minute activity — it is a five-minute activity and an hour. Everything below is chosen to avoid that.
1. The exit ticket, but one question
In the last three minutes, one question about the single most important thing in the lesson. Not a summary, not "what did you learn" — one specific thing you need to know they can do.
The discipline is in the "one". An exit ticket with four questions is homework, gets rushed, and tells you less than the single question would have.
2. The hinge question
A multiple-choice question placed at the exact point where the lesson changes direction, where each wrong option corresponds to a specific misconception. The distribution of wrong answers is the data, not the score.
If a third of the room picked option C, you do not just know they are wrong — you know which wrong thing they believe, and you can address that instead of re-teaching the whole thing. Hinge questions are slow to write and enormously worth it, because they get reused every year.
3. The confidence poll
Ask the question, then ask how confident they are. The interesting cell is confident and wrong — that is a misconception that will survive normal re-teaching, because the student is not looking for a correction.
Confident-and-wrong is the single highest-value signal in formative assessment and almost nobody collects it.
4. The no-names "did that make sense?"
Ask it with names attached and everyone says yes. Ask it as a poll where the room only sees totals and you get the real number. Four hands out of thirty going up is a different lesson from four votes out of thirty, because the second number is honest.
Use it sparingly — it works because it is rare, and because you visibly act on the answer. Ask it every lesson and it becomes furniture. How to word it, and which questions should never be anonymous, is covered in class polls that get honest answers.
5. Two truths and a lie, about the content
Three statements about what you just taught; one is false. Students find the lie. It takes thirty seconds to answer and forces them to evaluate rather than recall, which is a different and more useful thing.
It also survives being reused: change one statement and it is a new question.
6. Fill in the blank, mid-explanation
Stop in the middle of a worked example and have the class supply the next step. The fill-in-the-blank format is ideal here because you can accept several phrasings of the same step, and because a partly-right answer scores partly right.
7. The one-sentence summary
"Explain today's idea in one sentence, to someone who missed the lesson." This is a written question and normally expensive to mark — unless you grade it on keyword coverage, listing the three ideas that must be present and accepting any answer that has them.
Suddenly a genuinely open question costs you nothing to mark, and you find out who can compress the idea and who can only repeat it.
8. Predict, then check
Before a demonstration or a result, everyone commits to a prediction. Then you show what actually happens. Committing first is the whole mechanism: a student who predicted wrong and watched themselves be wrong remembers it, whereas a student who was told the answer does not.
9. The misconception hunt
Show a worked solution with a deliberate error and ask where it goes wrong. Finding a mistake in someone else's reasoning is much harder than following correct reasoning, and it maps directly onto checking their own work.
A useful trick: use a mistake you actually saw in last week's marking. It is more realistic than anything you would invent, and the students who made it recognise it.
10. The retrieval starter
Three questions at the start of the lesson, all from previous lessons rather than today's. It takes four minutes, it is the best-evidenced thing on this list for actual retention, and it doubles as a register of what has decayed since you taught it.
Keep a running pool and pull from further back over time — last lesson, last month, last term.
Making these cheap enough to actually do
Every idea above dies if it creates marking. Three things keep the cost near zero:
- Let the objective questions mark themselves. Hinge questions, predictions, two-truths-and-a-lie and retrieval starters are all multiple choice or true/false. They should be scored the instant they are submitted and never touched again.
- Grade open questions on coverage, not on wording. A one-sentence summary judged on whether it contains the required ideas is auto-markable. Judged on whether it matches your sentence, it isn't.
- Only look at the ambiguous ones. A tool that hands you the four answers it could not settle, instead of marking them wrong, turns a thirty-script pile into a two-minute read.
The honest test of a formative check: if the results arrive after you have already taught the next lesson, it was not formative. It was just marking.
Where to start
Pick two. The retrieval starter, because the evidence for it is strong and it costs four minutes; and the confidence poll, because confident-and-wrong is information you currently do not have about your class.
Run both for a fortnight before adding anything else. Most attempts at formative assessment fail from doing ten things badly rather than two things consistently.
Quizoma supports all of these formats — including polls and keyword-graded written answers — and marks the objective ones instantly. See how it works.