In Roediger & Karpicke's 2006 study, what did students do, and what happened on the immediate test?

Students who repeatedly re-studied scored HIGHER on a test taken 5 minutes later than students who had practiced retrieval — a misleading short-term win.

Roediger & Karpicke (2006, Psychological Science) had students read a prose passage, then either re-study it repeatedly (study-study-study-study) or study once then test themselves repeatedly with no feedback (study-test-test-test).

Tested 5 minutes later, the re-study group looked better: roughly 81–83% vs 71%. This is the illusion of competence — re-reading feels fluent, so it flatters your short-term recall. If you only ever measure learning immediately, you'll pick the weaker method.