The learner gets the next question right.
The correction is still open beside them.
What exactly did we measure?
Feedback contamination occurs when support, hints, answer exposure or corrective cues remain inside a later performance strongly enough that the result no longer cleanly shows what the learner can do independently.
This is the fourth pillar in the second Feedback quartet beneath How Feedback Works. It owns the measurement problem created by helping: when does useful support make the next result impossible to interpret?
Quick Read
Feedback is supposed to help performance change, so some support during learning is desirable. The problem appears when that support remains present during the very attempt used to judge whether the learner has become independent. A worked answer beside the question, a teacher’s leading hint, a highlighted error location or an identical repeated item can all make performance look stronger than the underlying capability. Strong feedback systems therefore separate learning attempts from receipt attempts. During learning, support can be generous. During verification, reduce the support, change the surface and observe whether the learner can reconstruct the correction without answer leakage.
Learning Mode and Receipt Mode Are Different
In learning mode, the teacher can explain, model, hint and prompt.
In receipt mode, the question changes:
What can the learner now do without the same support?
Answer Leakage Produces False Confidence
The student corrects a question while reading the worked solution.
That proves the learner can follow the correction.
It does not yet prove independent reconstruction.
Identical Retesting Can Be Contaminated by Surface Memory
A student gets the original question right immediately after correction.
They may remember the answer rather than the mechanism.
Use a parallel item or changed representation when the aim is to test the repaired capability.
Hints Change the Meaning of the Score
With a hint, the score may represent capability under support.
Without a hint, it may represent independent capability.
Both can be useful measurements. They should not be confused.
Fading Creates Cleaner Receipts
The third sibling, Feedback Fading, owns the gradual withdrawal of correction. Contamination is one reason fading matters: independence becomes visible only after enough support has left the measurement.
Retesting Still Owns the Verification Loop
Feedback Retesting owns whether repair survives another attempt. Feedback Contamination adds one safeguard: make sure the retest itself has not already supplied the answer route.
A Practical Contamination Check
TARGET: infer meaning from context LEARNING ATTEMPT: teacher highlights clue words CORRECTION: explain how clue changes interpretation CONTAMINATED RETEST: same paragraph, same highlights CLEANER RECEIPT: new paragraph, no highlights SECOND RECEIPT: delayed question with different vocabulary INTERPRETATION: independent inference route visible
Assessment Boundary
How Assessment Works owns what a performance sample justifies broadly. Feedback Contamination stays narrower: it asks whether the corrective intervention itself altered the sample so much that we can no longer infer independent change cleanly.
Final Thought: Help Freely, Then Measure Honestly
Feedback should be generous while the learner is repairing the route and disciplined when the system asks whether the route now belongs to the learner.
FEEDBACK · SECOND FOUR PILLAR LEGS
Return to How Feedback Works, or continue through Feedback Conflict, Feedback Source Calibration and Feedback Fading.