Using live quizzes for formative assessment

Formative assessment asks one question: does the class understand this well enough for me to move on? A live quiz answers it faster than anything else available in a classroom, and it answers it in front of everyone, which is both its advantage and the thing that makes teachers wary of it.

This covers how to use one without it feeling like a test.

Separate it from grading, visibly

The moment students believe a mid-lesson question is graded, the answers stop being diagnostic. People guess safe, copy neighbours, and go quiet when unsure, which removes exactly the signal you needed.

Say it explicitly, more than once, and make the settings match the claim. If it is not graded, turn scoring off entirely. A leaderboard is a grade with a different name, and a class that has seen you use one will assume every future question counts.

Ask the question that separates understanding from recognition

A question students can answer by recognising the phrasing from your slides tells you they were awake, not that they understood.

Change the surface and keep the structure. If you taught the concept with one worked example, ask about a different one. If you taught a definition, ask them to identify which of four situations is an instance of it. Recognition collapses the moment the wording moves.

The most useful question type here is one where a common misconception is an available option. When forty percent pick it, you have found the thing to re-teach, and you have found it in ninety seconds.

Read the distribution, not the score

The chart matters more than who got it right.

Split roughly evenly between two options usually means the question is ambiguous or the two ideas were taught too close together. Re-read it aloud before assuming the class is confused.

One wrong option dominating is the useful case. That is a shared misconception, and it is worth stopping for.

A wide scatter across all options generally means guessing, which means the material has not landed at all and moving on is not an option.

Near-unanimous and correct means you can move faster than planned, which is a finding too.

Do something with it in the same lesson

The whole value of formative assessment is that it changes what happens next. If the answer to a bad result is to continue as planned, the exercise was decorative.

Re-teach the specific thing, not the whole topic. Three minutes on the misconception the chart identified is worth more than repeating the original explanation, which already failed for this group.

Then ask a second question on the same idea, worded differently. Students read a re-ask as fair rather than punitive when you tell them it is coming, and it gives you a clean measurement of whether the re-teaching worked instead of an inference from atmosphere.

Anonymity, and its limits

Anonymous responses get you honest ones, particularly from students who track their standing relative to peers. For anything where being wrong is socially costly, anonymous is the right default.

The limit is that you lose the ability to follow up with an individual. If you need to know which three students are lost, a quiz is the wrong instrument and you want the work in front of you instead.

In a class under about ten, anonymity is thin regardless of the setting: everyone can work out who the two outlying answers belong to. Ask about the material rather than about the person at that size.

A shape that works in a 50-minute lesson

Two questions about ten minutes in, checking the prerequisite you assumed they had. This catches the lesson that was doomed before it started.

Two more after the main explanation, both on the central idea, one of them with a misconception as an option. Stop and re-teach if the chart says so.

One at the end that you also plan to ask at the start of the next lesson. The comparison between the two is the most honest retention data you will get, and it costs ninety seconds.

Assigning it instead of running it live

Some checks work better as homework than in the room: the ones where you want considered answers rather than fast ones, and the ones where a student being wrong in public would stop them answering honestly.

A self-paced assignment gives you the same distribution without the performance element. You lose the moment where the class sees that half the room shares their confusion, which is often worth more than the data. Use live when the shared realisation matters and assigned when the individual answer does.

What this will not tell you

Whether students can apply the idea to an unfamiliar problem. Multiple choice measures recognition and discrimination well and transfer badly.

Whether the quiet ones understood. A correct answer from a student who guessed between two options looks identical to a confident one.

Whether it will still be there next week. That needs the same question again, later, which is why the end-of-lesson question is worth re-asking rather than retiring.