Assessment for learning strategies that survive a real lesson

Assessment for learning, AfL if that is what the department calls it, has a plain meaning. You find out whether they have got the thing you just taught, and you change what you do next because of what you found. If nothing changes, it was not assessment for learning. It was a ritual.

This is not a history of the term, and it is not the piece about why a nod is worthless. That argument is in how you know who understood. This is the short list worth keeping on a day when you teach five lessons and eat lunch at a desk. The rest of the notes are on the blog.

Start from the decision, not the activity

Before you pick a strategy, write the decision the check is supposed to change. "If most of them still think the current gets used up, we do not start resistance." "If they can already embed a quotation, we skip the model and go to the extract." A strategy with no decision attached becomes a poster.

There are only a few decisions that matter in a lesson. Carry on. One more example, taught a different way. Stop, because the foundation is wrong and the next task will practise the mistake. Send three pupils the harder problem and sit with six. Everything else is detail.

Success criteria they can actually use

A learning objective on the board is not a success criterion. "Understand meanders" cannot be checked by a pupil, because they cannot see their own understanding. "Explain why the outside bend is deeper, without using the diagram" can.

Keep it to two or three lines, in the words you would accept in a book. Then make them use it on one piece of work, not admire it. They do a short answer, point at the line they think they have met, and you look at five books and say which line is missing in the room. If you never look, the criteria were decoration. Twelve lines will all get highlighted, and you will learn nothing.

A hinge, once, in the middle

Somewhere in the lesson there is a point where going on would be a mistake if a chunk of the class has the wrong idea. Plan one question for that point. Every pupil answers. Each wrong option is a mistake you have really seen, not a joke answer.

That is a hinge question, and it earns its place because the lesson can still bend. An exit ticket cannot bend the lesson you have already finished. A starter cannot tell you whether today's explanation worked. The hinge is the AfL strategy that sits in the middle on purpose. One is enough. Three hinges and you no longer have a lesson. You have a quiz with interruptions.

Every pupil answers something

Hands-up AfL is AfL for six pupils. You already knew those six. The strategies that change a lesson are the ones where the quiet pupil also has to commit: a word on a mini whiteboard, a sentence in the margin, a private answer on a device, a name you choose after everyone has had time to think.

A menu of those moves, and what each one hides, is in checking for understanding. The principle is smaller than the menu. If you cannot name pupils who have not got it, you have not checked. You have had a conversation with the willing.

"Any questions?" fails this test. So does "thumbs up if you're happy". So does traffic-light cards where pupils rate themselves. Self-report is not evidence. The pupil who most needs you is the one most likely to show you green, because red is public. Use a traffic light only if it records an answer you can see, not a feeling.

Feedback is the next task

A comment is not feedback until the pupil does something because of it. "Add more detail" is a wish. "Rewrite the last sentence so the quote proves the point about control, not the whole plot" is a task. In the room, the whole-class version is the same idea. You do not write thirty comments. You name the error you saw in the sample, show a weak answer and a better one without naming the pupil, and set two minutes to fix that one thing.

If the sample of five all made the same error, stop and reteach. Writing "see me" thirty times is a slow way to do a whole-class correction. The check was for you. The mark book can wait.

Sampling, properly

"I've had a look" is not a sample. Pick five names before the lesson, from the pupils you are least sure about, not the five neatest books on the top of the pile. Read those live. If three of the five share a mistake, assume it is in the room and check one more from the back corner before you decide.

Live sampling beats taking the books in, because you can still say "stop, look at this". A set taken home on Tuesday tells you about Tuesday on Thursday. That can be assessment of learning. It is a weak form of assessment for learning.

What to drop without guilt

• Lolly sticks as the whole system. Useful for making yourself pick someone other than the usual hands. Useless if the stick is the only information you gather. One name is not a class.
• Peer assessment with no criteria. "Give your partner a star and a wish" produces kind comments and very little change. Peer assessment works when the partner is checking one line you named, and you still sample the results.
• Mini plenaries that are a second explanation from you. If you talk again, you have not assessed anything. You have repeated yourself.
• Confidence ratings, exit tweets, and word clouds of "today I learnt". They document that the lesson happened. They do not show you the pupil who cannot do it.

A shape you can repeat tomorrow

Objective in language a pupil could fail. Ten to fifteen minutes of teaching. One hinge, everyone answers, you either go on or you do not. Practice while you sample five books. One correction aimed at the error you actually saw. A single question at the end if you need names for tomorrow.

That is the whole strategy. It does not need a new app every half term, and it does not need a worksheet called "AfL". It needs a question you wrote the night before, and the nerve to stop when the answers say stop.

When you do want the names, and you do not want the pile, type the topic into Get It?. It drafts the check, you edit it, they answer in their own words on any device with a code, and you see who has got it, who is unsure, and who needs help. Free for a limited time, no card, no pupil accounts. Use it for the decision. Do not use it as a second test.

Questions people ask

Try it on one class.

One lesson. One question. Two minutes. If it doesn't tell you something you didn't already know, walk away.

Try it free — no card

Students need no account. They just need the code.