The problem: recognition passes for recall
A GCSE maths student can read a worked solution, follow every line, and still fail the same question a week later. The cause is cue dependency: the printed method supplies the first step, so the brain never has to retrieve it. A duffspin set removes that cue by asking for one answer at a time, then checking it immediately. The aim is not speed; it is to make the missing step visible while the student can still fix it.
Example: a Year 10 student revising solving equations may correctly identify inverse operations on a worksheet but stall on a blank page when the equation includes a fraction. The takeaway is that revision needs to test production, not familiarity.
The cause: error loops hide behind neat notes
Neat notes create a second problem: they record the correct method after the learner has already corrected it. The original error disappears. A duffspin set keeps a short log of the first wrong move, the correction, and a near-identical question answered two minutes later. That log turns a vague feeling of ‘I know this’ into a specific target, such as ‘I drop the negative when multiplying inequalities’.
In a tutoring session, the log also gives the tutor something concrete to probe. Instead of asking whether the topic feels better, the tutor can ask the student to explain the exact step that changed between attempt one and attempt two.
How to run a duffspin set
A useful duffspin set is small enough to finish in five minutes and narrow enough to repeat the same decision. Use one topic, one page, and one visible timer.
- Choose one decision. Pick a single step, such as expanding a double bracket or choosing sine versus cosine.
- Write three questions. Keep them short and varied enough that the student cannot copy the previous answer.
- Answer without notes. The first attempt is diagnostic, not a performance.
- Check and name the error. The student says the wrong move aloud, then rewrites the correct line.
- Repeat one near-identical question. This closes the loop while the correction is still active.
For a tutor-led version, duffspin works best when the adult holds the questions back until the student has committed to an answer. That small delay prevents the learner from reading the method instead of retrieving it.
What to do with the errors
Errors are the product of a duffspin set, not a side effect. Sort each mistake into one of three buckets: a fact gap, a process gap, or a reading gap. A fact gap needs a flashcard; a process gap needs a slower worked example; a reading gap needs the question rephrased. The solution is not more questions on the same night. It is a second duffspin set 24 to 48 hours later, using the same decision with different numbers.
Stop a set when the student can answer one fresh question correctly and explain why the wrong move was wrong. That condition is stricter than a correct answer and more useful than a tick on a revision timetable.