Measuring actual learning versus feeling of learning in response to being actively engaged in the classroom

Two large-enrollment introductory college physics courses at Harvard; students randomly assigned within course to active or passive instruction for matched content

UNDERGRADUATE introductory physics students — outside this archive's age range (4-18).

Recorded because a founder will be handed this paper, and because its finding is the most useful single caution in the whole active-learning literature. Students within the same course were RANDOMLY ASSIGNED to receive identical content and identical handouts either through active instruction following disciplinary best practice, or through a lecture delivered by an experienced, highly-rated instructor. The instructor made no attempt to persuade students of the merits of either method. Both actual learning (a test of learning) and perceived learning (self-report) were measured. The design is unusually clean for this literature — same content, same materials, randomised, crossed instructors — which is why it survives as a B despite being a single-institution study.

Students in the active classrooms LEARNED MORE, as measured by the test of learning, but their SELF-REPORTED perception of how much they had learned was LOWER than that of students in the passive lecture condition. The authors attribute the gap in part to the greater cognitive effort active learning demands: fluent, well-delivered lecture feels like learning; effortful engagement feels like struggle. Two implications carry into K-12 even though the study does not: (1) student and teacher satisfaction ratings are an actively MISLEADING proxy for learning, and can push a school towards the inferior method — "a superstar lecturer could create such a positive feeling of learning that students" prefer it; (2) the same logic explains why science programmes that reliably raise pupil ENJOYMENT and SELF-EFFICACY (the EEF Thinking, Doing, Talking Science trials moved both in the right direction) can move attainment not at all. Effect sizes were not extracted from the paper and are not recorded here.

Low — within-course random assignment; the exposure is a single institution and a highly selected student body.

The counterpart result to Freeman et al. 2014 (active learning raises measured performance). The perception-versus-learning dissociation itself has not been directly replicated in a school-age science setting so far as this archive has checked.