The effectiveness of test-enhanced learning depends on trait test anxiety and working-memory capacity.

Despite being viewed as a better way to enhance learning than repeated study, it has not been clear whether repeated testing is equally effective for students with a wide range of cognitive abilities. The current study examined whether test-enhanced learning would be equally beneficial to participants with varied working-memory capacity (WMC) and trait test anxiety (TA). Chinese-English bilingual undergraduates in Hong Kong were recruited as participants. They acquired Swahili-English word pairs (half via repeated study and half via repeated testing) and performed a delayed cued-recall test for all pairs about one week after the acquisition phase. Their WMC and TA were estimated by Unsworth, Heitz, Schrock, and Engle's (2005) operation-span task and the Chinese version of Spielberger's (1980) Test Anxiety Inventory, respectively. We replicated the typical testing effect: Participants performed better for pairs in the repeated-testing condition than those in the repeated-study condition. Regression analyses showed that, (a) relative to other participants, those with lower WMC and higher TA made more intralist intrusion errors (i.e., recalling a wrong English translation to a Swahili word cue) during the acquisition phase, and (b) the testing effect was negatively correlated with TA for participants with lower WMC, but was not correlated with TA for participants with higher WMC. This demonstrates a boundary condition for the use of test-enhanced learning. Implications of these findings for theories of the testing effect (e.g., Pyc & Rawson's, 2010, mediator-effectiveness hypothesis) and their application in classroom settings are discussed.