Students are often told to check their work.
That advice sounds complete.
It is not.
Because checking only works if the student knows what they are searching for.
A vague reread can create the feeling of diligence while missing the exact error that matters.
Examination Craft therefore treats checking as a search problem.
Quick Read
Effective checking is targeted error search. Instead of rereading every answer with the same method and attention, students identify high-risk locations and likely error types, then use checks capable of detecting those errors. Good checking is selective, independent where possible, and informed by the student’s own error history. The aim is not to feel more certain. It is to find correctable mistakes before submission.
Why Ordinary Rereading Often Feels Better Than It Works
The student has just produced the answer.
They know what they intended to write.
When they reread it immediately, the intended meaning can fill gaps that are not actually present on the page.
The same route that created a mathematical error may reproduce the same error during checking.
The same interpretation that caused task drift may still feel persuasive on the second read.
So “look again” is not enough.
A useful check needs a target.
Checking Begins with a Hypothesis
Ask:
If this answer is wrong, what is the most likely reason?
Then search for that failure.
For one student, the likely error may be copied numbers.
For another, negative signs.
For another, answering a comparison as two separate descriptions.
For another, leaving the causal mechanism implicit.
The stronger the student’s error model, the more efficient the search becomes.
The Four Layers of Checking
1. Completion
Did I answer every required part?
Did I miss a page, sub-part, choice or instruction?
2. Constraint
Does the answer obey the conditions of the question?
Correct unit?
Required accuracy?
Specified source?
Requested number of points?
3. Reasoning
Does the route actually support the result?
Is a step missing?
Does the evidence support the claim?
Does the mechanism connect cause and consequence?
4. Plausibility
Does the answer make sense in the world of the question?
Is the magnitude reasonable?
Does the conclusion contradict the data?
Did the interpretation become stronger than the evidence permits?
Use a Different Check from the Method That Created the Answer
Independent checks are powerful because they are less likely to repeat the same route failure.
In Mathematics, substitute the answer back, estimate the size, use an alternative relation or check units where appropriate.
In Science, test whether the explanation predicts the observation actually given.
In English and Humanities, compare the final sentence with the command word and required object: did the response actually perform the requested action?
The exact checking methods belong to each subject.
The cross-subject principle is simple:
Do not merely repeat the production process and call it verification.
Your Personal Error Signature
Checking becomes more valuable when it is personalised.
Take several marked papers and classify preventable errors.
- misread command;
- missed constraint;
- copied value incorrectly;
- sign or arithmetic slip;
- unit error;
- incomplete explanation;
- evidence without link;
- question left partially unanswered;
- answer changed without evidence;
- over-answering that caused later rushing.
Count what actually recurs.
That becomes the student’s search priority.
This is more useful than copying somebody else’s generic checklist.
Confidence Is Evidence, but Weak Evidence
Confidence can help prioritise checking.
An answer that felt uncertain deserves attention.
But confidence is not proof of correctness.
Research on metacognitive calibration shows that students vary in how accurately their confidence matches actual performance. Error detection depends partly on this monitoring accuracy.
So combine confidence with structural risk.
A multi-step answer can deserve checking even if it felt smooth.
Do Not Change an Answer Merely Because Anxiety Appeared
Checking sometimes creates doubt where none existed before.
The student sees the answer again and thinks:
“Maybe I should change it.”
The better rule is evidence-based:
Change an answer when the check produces a reason, not merely a feeling.
A discovered constraint.
A contradiction.
A calculation error.
Better evidence.
A clearer interpretation.
That is a justified change.
Checking Under Scarce Time
If time is abundant, checking can be broad.
If time is scarce, it must be selective.
Prioritise:
- blank or incomplete required work;
- answers where constraints are easy to verify;
- questions with several fragile steps;
- answers marked uncertain during execution;
- your recurring error signature.
This is why The Last Fifteen Minutes and checking belong together.
A Practice Drill: Error Hunt
Take a completed paper before marking it.
Give the student a limited checking window.
Before checking, require three predictions:
- Where is an error most likely?
- What type of error is it likely to be?
- What checking method could detect it?
Then compare detected errors with the marker’s final findings.
Over time, the student improves not merely at solving questions, but at predicting where their own performance is fragile.
What Parents and Tutors Should Diagnose
When a student says “I checked already,” ask what the check actually did.
- What error were you searching for?
- Which questions did you prioritise?
- Did you use an independent check?
- Did you discover anything?
- Do your recurring mistakes match what you chose to inspect?
A checking routine that never finds errors despite recurring preventable losses is not yet an effective checking system.
Where This Fits in Examination Craft
The operational sequence now reaches verification:
interpret → decompose → initiate → execute → monitor → allocate → verify.
Checking Is a Search Problem owns the verification job.
The next estate should move into error signatures, answer-changing, blank-mind recovery and post-paper calibration.
Evidence and further reading
Recent educational research links metacognitive calibration with error detection and self-monitoring. That supports a cautious Examination Craft principle: checking improves when students learn to predict where their own answers are vulnerable and compare confidence with actual performance.
- npj Science of Learning — metacognitive monitoring and biology error detection
- Frontiers in Education — student metacognitive monitoring and error detection
- Journal of Chemical Education — monitoring exam performance and calibration
The useful sentence
Do not check by staring harder at everything. Search deliberately for the errors you are actually likely to make.