I recall reading a paper a few years ago (though I think it was in physics) on a class where they decided to repeat a midterm test question on the final exam and students performed noticeably worse on the final exam than on the midterm (in fact only one person's grade noticeably increased I think).

Can anyone help jog my memory with which paper this might have been? Alternatively, if you know of another paper that replicated this effect in mathematics and could let me know that would work as well.

Thanks!