ABSTRACT
The effect of self-explanation on learning has rarely been tested experimentally, using sufficient sample sizes and active control conditions. In this preregistered study (N = 208), a 2 × 2 mixed design was implemented to test whether (1) factual information is better retrieved after self-explanation versus re-reading, (2) memory depends on retention interval (immediate vs. 2-week delay), (3) learning strategy and retention interval interact, and (4) the number of self-explanations predicts performance. Participants learnt fictional facts under laboratory conditions. An ANOVA revealed a large effect of retention interval, but no effect of learning strategy and no interaction. A multiple regression revealed that—within correct self-explanations—the number of inferences predicted factual test performance, but not the number of paraphrases. Thus, although producing correct inferences during self-explanation was associated with better performance, self-explanation itself had no effect on factual learning compared to re-reading, neither immediately nor after a delay.