Long-Term Hypercorrection, Return Errors, and the Transfer of Learning in the Classroom

We report three classroom studies on the hypercorrection effect—misconceptions that learners are more confident about are more likely to be corrected than misconceptions they are less confident about. Studies 1–2 test whether corrected misconceptions return after an extended delay: Students completed a pretest over course concepts, rated their confidence in the correctness of each response, and were shown correct-answer feedback. Students then completed an immediate (identical to the pretest) and delayed posttest with novel questions that tested identical concepts as the pretest. Both studies found a hypercorrection effect on the immediate posttest, but about 20% of misconceptions that were corrected returned on the delayed posttest, particularly those that students were highly confident about. Study 3 only included an immediate posttest with novel questions that covered analogous concepts as the pretest and found a hypercorrection effect, thus demonstrating that these corrected misconceptions reflect learning that extends beyond rote memorization.