Did you know that reading a beautiful worked solution can make a pupil feel confident without proving that the pupil could have produced the solution independently? The final answer looks familiar, the lines seem reasonable, and the learner nods. Yet the next question, with slightly different numbers, may still be impossible. The missing step is often a genuine attempt before seeing the model answer.
Committing to a prediction before reading a worked solution means writing down your proposed answer, the reasoning behind it and what you are uncertain about, before revealing the model steps. Compare your first attempt with the solution, diagnose the first invalid move, repair it, and then solve a genuinely different question independently. You are testing how you think, not just whether the printed answer looks right.
Practical summary: hide the solution; state the question in your own words; write a predicted answer or first step; give a reason and a confidence estimate; reveal the solution; identify the first difference; explain the correction; close the solution and test a fresh example. A mistake uncovered early is useful information, not a verdict on the learner.
Why predictions before answers are different from copying a solution
A worked example provides a sequence of valid moves. It can show what skilled reasoning looks like and relieve the learner of inventing every step from scratch. That makes it particularly helpful when a new skill is introduced. However, familiarity while reading is not the same evidence as an independently generated method.
The distinction is simple. While reading 3(x + 4) = 30 followed by x = 6, it is easy to agree. Before looking, a student must decide how multiplication and brackets interact, what to do to both sides, and how to check the result. Recording that decision leaves a visible trace of the learner’s understanding.
The Learning Scientists’ research-informed guidance explains that active retrieval can support later application and that feedback matters when responses are mistaken. Their student guide to retrieval practice recommends checking an attempted answer against notes afterwards. A pre-solution prediction can incorporate that logic, but this particular classroom routine should not be presented as a guaranteed stand-alone treatment or a replacement for explicit teaching.
Worked example: an algebra answer that looks plausible but fails its own test
Consider the fictional student Maya, practising this equation in a Secondary 1-style algebra lesson: 3(x + 4) = 30. The question asks for the value of x. Maya covers the worked example and writes the following prediction: ‘Divide by 3 to get 10, then add 4; therefore x = 14.’ She rates her confidence as 4 out of 5. This is an invented teaching example, not a report of a real student.
- Prediction recorded: x = 14, with a reason: ‘the bracket contains plus four, so I add four at the end.’
- Quick check: put x = 14 into the original expression.
3(14 + 4) = 3 × 18 = 54, not 30. Her predicted answer fails. - Correct method A: divide both sides by 3 to get
x + 4 = 10; subtract 4 from both sides to obtainx = 6. - Correct method B: expand
3(x + 4) = 3x + 12; then3x + 12 = 30,3x = 18, andx = 6. - Verification:
3(6 + 4) = 3 × 10 = 30. Both sides match, so 6 satisfies this equation.
Maya’s important misconception is not ‘bad at algebra’. She correctly divided 30 by 3. The error appeared in her interpretation of x + 4 = 10: to isolate x, she must undo adding four by subtracting four, rather than add four again. The first wrong step is therefore more informative than the final incorrect number.
A teacher can now ask: ‘If the inside of the bracket must be ten, should x be smaller than ten or larger?’ Because x plus four equals ten, x must be less than ten. This reasoning check works even before detailed arithmetic, and it becomes a reusable habit for new equations.
The five-column attempt log
A compact notebook record can prevent vague corrections such as ‘I wasn’t careful’. Use five columns labelled Question, My prediction, Why I thought so, First difference, New independent test. The entry is short enough to complete during a lesson yet specific enough to guide what happens next.
- Question:
3(x + 4) = 30, solve for x. - My prediction: x = 14; confidence 4/5.
- Why I thought so: I divided 30 by 3 and added four.
- First difference: the solution subtracts four to undo
+4. - New test: without opening the solution, solve
4(y + 3) = 36; sincey + 3 = 9,y = 6; check4(6 + 3) = 36.
Notice the independent test uses another coefficient and another bracket constant. It is not identical to the solved example. The learner must apply the meaning of an inverse operation, not copy the number six simply because it appeared above. In this case the independent answer also happens to be six, so a teacher should give a second transfer question with a different answer, for example 5(z − 2) = 30, which gives z − 2 = 6 and therefore z = 8. The substitution check is 5(8 − 2) = 30.
How to make the routine precise without making it stressful
Step 1: Choose the right moment
Early in a new topic, first show one fully explained example and allow guided practice. It would be unhelpful to demand unsupported predictions about an unfamiliar principle from a pupil who has not yet had the necessary teaching. As understanding develops, begin covering part of the solution, asking for the first step, and then covering the entire model.
Step 2: Record a genuine commitment
Ask the learner to write an answer, an equation, a diagram, or the first justified move before seeing the model. A verbal guess can work in a live conversation, but a written trace makes it easier to compare exactly what changed. For language learning, the commitment might be an edited sentence; for science, a prediction of which way a measured variable will move; for history, the claim supported by a particular source.
Step 3: Add reasons, not just confidence
‘I think it is 14’ records little about the misconception. ‘I divided by three and then added four’ identifies a specific move that can be tested. A confidence estimate from one to five helps learners notice when they are very sure but wrong. It is a self-reflection cue, not a calibrated scientific measurement of knowledge.
Step 4: Compare the first difference
Once the model is visible, read it as an explanation rather than a set of lines to reproduce. Ask which step first differs from the attempt, and why the model step is valid. Sometimes the learner’s different method is also correct; do not force everyone to copy identical algebra if two methods are mathematically equivalent.
Step 5: Close the model, attempt a transfer problem
Checking a solution is incomplete until the pupil can work without it. Choose an adjacent but not identical problem, delay the answer, and require a check. For long-term learning, revisit the idea later instead of relying only on the success of a correction made seconds after seeing the model.
Learner diagnostic: three possible explanations for the same wrong answer
Suppose a pupil writes 5(z − 2) = 30 then z = 4. That single wrong result is not a sufficient diagnosis. It could come from confusing the subtraction inside the bracket, an arithmetic slip, or a misunderstood inverse operation. Ask the student to explain the transformation of z − 2 = 6. Then supply a simpler check such as z − 2 = 4 or a visual number-line representation.
If the learner correctly solves the simpler equation but made a one-off copying error, the repair may be a written check routine. If they repeatedly subtract two to isolate z, teach inverse operations using balance and substitution. If they cannot follow any step, return to prerequisite arithmetic with signed numbers. The action should match the evidence; additional random worksheets may not.
Progressive practice: guided, partial, independent
- Guided: show a model for
2(a + 5) = 24, pause before the final line, and ask fora + 5 = 12anda = 7. Check2(7 + 5) = 24. - Partially guided: show only the starting equation
6(b − 1) = 42. Ask the student to predict whether b is larger or smaller than 7 before calculating. Sinceb − 1 = 7, b must be larger:b = 8. - Independent: present
4(c + 2) = 20without hints, pause for a full solution, and require substitution.c + 2 = 5, soc = 3, and4(3 + 2) = 20. - Delayed transfer: after a break, use
3(d − 4) = 15. The solution isd − 4 = 5, henced = 9. Ask for the first step and an independent verification before showing any answer.
Each stage changes the amount of help, not the purpose of the learning. A teacher should move forwards when the pupil can explain the transformation on several varied questions, and move backwards when the first step reveals a missing prerequisite. Faster completion alone is not proof of understanding.
What parents and teachers can observe
For students: preserve one original wrong line alongside its corrected explanation. You should be able to tell another learner why the correction is valid and apply it when the numbers change. For parents: ask ‘What did you think before you saw the answer?’ and ‘What would happen if we changed this number?’ rather than asking only for a mark. Praise a valid check and a thoughtful repair, not just confidence.
For teachers: choose a short hinge problem, collect the predictions, and group feedback by first misconception. If several learners made the same invalid inverse operation, reteach that move with diagrams and examples. If the class can solve identical items but fails fresh numbers, provide varied independent questions rather than more copied examples.
A small progress record could show how many of five fresh equations a student solved correctly without a displayed model, whether the student checked substitution, and whether an explanation named the inverse operation. This is an observable local learning indicator, not a promised score increase or a standardised research outcome.
Common misunderstandings to correct
- ‘If I guessed wrongly, the method failed.’ A wrong guess can be the most useful part if it exposes an exact misconception and is followed by accurate feedback.
- ‘I understand because the answer looks obvious now.’ Recognition while viewing a worked solution is weaker evidence than correctly generating and checking a fresh answer.
- ‘Only the final answer matters.’ A lucky correct number can conceal invalid reasoning, while a mostly correct method can contain a repairable arithmetic slip.
- ‘All worked examples should be hidden from beginners.’ New learners can benefit from explicit explanation and scaffolding; the point is to increase independent contribution gradually.
- ‘Confidence means correctness.’ Confidence should be compared with performance, not used in place of testing the equation.
Frequently asked questions
Should the pupil predict the whole answer or only the first step?
Choose the smallest useful commitment that the pupil can realistically attempt. For a new method, a predicted first step plus a reason is enough. For an established method, request a full solution, an explanation and a check.
Can this technique help with English or science?
Yes, if the prediction is specific and checkable. A pupil can propose a subject–verb correction, predict which sample will have a higher measured value, or select which sentence actually supports a claim. The feedback must address the relevant language or science principle; not every classroom question has a single numerical answer.
What if the pupil copies the model after making a prediction?
Copying may produce tidy notes but is not the final check. Hide the model again and use a fresh question with slightly different conditions. Listen to the reasoning and ask the learner to verify rather than replicate it.
Do we need to give a confidence score every time?
No. It is an optional, lightweight prompt that helps a learner notice overconfidence. If scoring makes a student anxious or diverts attention from the reasoning, omit it.
Does one correct new answer prove mastery?
No. More varied independent examples and later retrieval provide stronger evidence than one success immediately after correction. Watch for consistent explanations across contexts.
A helpful final habit: think, reveal, explain, transfer
The aim is not to turn every solution into a test. It is to make learning visible at the moment when a student would otherwise be tempted to accept an answer without doing any thinking. If the pupil can identify the first wrong move, understand why it was wrong and solve a different problem alone, the model has become a tool for independent capability.
Continue with eduKate’s verified guide to testing recall without notes and the worked explanation about negative exponents. The Core Aim of Education hub connects these skills with truthful capability. The immutable Clementi Mathematics learning reference demonstrates progressive teaching, diagnosis and independent checking.