Marking tells you who got it wrong after the lesson has ended. A check during the lesson tells you the same thing while you can still reteach the bit that broke. That is the only version of the information you can act on, and it is the only job the questions below are for.
Ask a class whether everyone follows and you learn who is willing to say no in front of their friends. The quiet middle stays quiet. The confident student answers first and sets the reading of the room, and a misconception plenty of them hold sails through unnoticed until it turns up in an exam months later. Answering privately, all at the same time, removes both problems. Be clear about the trade, though. Answers here are anonymous by design: there is no proctoring, no identity verification, and nothing stops a student answering twice or answering for a friend. That makes this a good instrument for a low-stakes check and a bad one for anything that carries a grade.
Six questions worked through on one topic — averages and correlation — so the sequence reads as a real lesson. Swap the subject and keep the order: retrieval practice first while it is cold, the misconception check second, confidence last.
Cold recall from the previous lesson, before any new material. Retrieving an answer strengthens it far more than rereading it, and you find out what survived the week.
Deliberately not the quiz type. Scoring makes students answer what they think you want; unscored, you get the wrong answers, and the wrong answers are the whole point of asking.
The choices told you how many hold the misconception. Only their own words tell you which misconception it is, and whether the right answer was reasoning or a lucky guess.
Confidence read against performance. High scores with low confidence means practise; good confidence with poor answers is the dangerous combination, and neither shows up unless you ask both.
Vocabulary is often the actual blockage. Whatever clumps in the middle is the term to define again at the start of next lesson, and it costs the class ten seconds to tell you.
Ends the check with a decision rather than a mood, and the class has just given you the evidence for it.
A distractor invented on the way to the lesson diagnoses nothing. The useful wrong options are the mistakes last year's class actually made, written in the words they used, so the count against each one names the misconception.
When a diagnostic question splits the room, do not give the answer. Show the split, have them defend their choice to the person next to them, then run the same question again. The movement between the two charts is the evidence that the discussion worked.
Use the quiz type where you want effort and a bit of pace, and plain multiple choice where you want the truth. The moment an answer is marked right or wrong, some of the class starts guessing at your intent instead of reporting theirs.
There is no sign-in, no name on an answer and nothing verifying who is holding the phone. That is deliberate, and it is exactly why students answer honestly — but it means these results are evidence about the class, never a record about a student.
A check that changes nothing in the next ten minutes trains the class to answer carelessly. Reteach the split, or say out loud what you are moving to next lesson because of it.
Every question above, ready to edit into your own topic. Put the join code on the board and students answer from whatever is in their pocket, without an account.
Poll a hall of three hundred and find out who is lost before you move on
Discussion questions that start with everyone already committed to a position
Hear from all thirty, not the four with their hands up
Give remote students the same way in as the people in the room