You hand back a test. Half the class immediately flips to the wrong answers and starts copying the correct one from a classmate, word for word, to get half-credit back. They walk away with a slightly better grade and exactly the same misunderstanding they walked in with.
Test corrections are supposed to be a second learning opportunity. Most of the time they're just a grade-recovery transaction, because the policy only checks whether the final answer is now correct, not whether the student understands why.
Why copy-the-right-answer corrections fail
The whole value of a correction is supposed to be: go back into the moment where your thinking went wrong, and fix the thinking, not just the answer. A policy that only checks the final answer skips that step entirely. The student learns "get the answer from somewhere" instead of "find where my reasoning broke."
A structure that forces the actual re-learning
For each missed question, require three things, not one:
1. What I originally thought, and why. A sentence in their own words about their original (wrong) reasoning. This is the part students want to skip — it's also the part that does the actual diagnostic work, for them and for you.
2. Where specifically the reasoning broke. Not "I made a mistake" — the specific step. "I forgot to balance the equation before calculating moles" is useful. "I got it wrong" is not.
3. A new problem, not the same one. Solve a similar but different problem correctly, not just rewrite the original with the right number swapped in. If they can only fix the exact problem in front of them, they haven't relearned the concept — they've memorized one answer.
Partial credit is earned back based on quality of the explanation and new problem, not just whether the original answer is now correct.
Making this manageable to grade
This sounds like more work than checking a copied answer, and it is — unless you build the structure once and reuse it. Set up a standard correction template with these three prompts built in, so students know the expectation before they even get their test back, and you're scanning for quality of reasoning rather than reading free-form paragraphs each time.
Our [rubric designer tool](/grading-rubric) can build a scoring structure specifically for correction-quality (reasoning explanation + new problem accuracy) rather than a generic rubric, and our [quiz and assessment builder](/quiz-generator) can generate the similar-but-different follow-up problems for each missed question type automatically, so you're not hand-writing a parallel version of every question a student got wrong.
What changes when you do this
Test corrections stop being a grade-recovery loophole and start being genuinely the second-best moment to teach a concept, right after the original lesson, while the student still remembers exactly where they got confused. That's a real teaching opportunity most correction policies throw away for the sake of a faster grading pass.
