A Meta-Analysis of the Effectiveness and Boundary Conditions of Retrieval Practice, Distributed Practice, and Interleaved Practice in Authentic School Classrooms

Abstract While retrieval practice, distributed practice, and interleaved practice are considered three promising strategies for how, when, and in what sequence to practice after initial instruction or self-study, little is known about the average effects of these strategies in authentic classrooms in primary and secondary education and their moderators. Therefore, in the present study, we investigated the effectiveness and boundary conditions of retrieval, distributed, and interleaved practice in such classrooms. We did this using a three-level meta-analysis comprising 233 effect sizes from 53 studies (published 2006–2024) including 7,303 participants. We found (a) an average testing effect of g = 0.37, moderated by type of knowledge, posttest format, format matching between practice and test phase, and type of learning outcome, (b) an average spacing effect of g = 0.20, moderated by type of comparison condition, retention interval, the ISI/RI ratio (i.e., the interstudy interval expressed as proportion of the retention interval), and posttest format, and (c) an average interleaving effect of g = 0.40, moderated by provision and timing of feedback. Average effects suggest that the effectiveness of the three strategies extends to authentic school classrooms. Observed moderator patterns suggest that the testing, spacing, and interleaving effects only occur reliably if their underlying cognitive processes are elicited during practice. This, in turn, implies that we need to shift from a one-size-fits-all instructional approach to a context-dependent instructional approach.