The Magnitude of the Testing Effect Is Independent of Retrieval Practice Performance

Practicing retrieval is a potent learning enhancer. Theoretical accounts of the testing effect generally suggest that the magnitude of the testing effect is dependent on retrieval practice performance, such that conditions that promote better retrieval practice performance should result in a greater testing effect. Empirical evidence, however, has been mixed. Although some studies showed a positive association between retrieval practice performance and the testing effect, others have shown either no relation or the reverse. In the present study, we experimentally manipulated retrieval practice performance using a retrieval-based response deadline manipulation and an encoding-based study trial manipulation. Across six experiments, the magnitude of the testing effect was independent of retrieval practice performance. However, when we aggregated the data across the experiments, participants with superior retrieval practice performance showed a greater testing effect-an individual difference. This dissociation between experimental and correlational outcomes suggests that the positive relation between retrieval practice performance and the testing effect is not causal, and indeed, simulation data showed that the correlation between retrieval practice performance and testing effect was an artifact. We discuss the challenges these findings present to existing accounts of the testing effect.