Theses Doctoral

An Investigation of Individual Differences in the Testing Effect

Kern, Michael B.

Retrieving information from memory via a test has long been considered primarily a method of evaluating knowledge. However, a growing body of experimental research has demonstrated that taking tests also helps to strengthen memory. In both experimental and naturalistic settings, research has established that, compared to rereading or reviewing, practicing retrieval through test taking leads to poorer performance on immediate tests but superior and more durable long term memory on delayed tests. The interaction between time of test and method of study is called the testing effect. To date, very few attempts have been made to determine what cognitive variables might affect the existence or size of the testing effect, and none have done so using correlational research designs. This dissertation examined the effects of language comprehension, background knowledge, and metacognition on individuals immediate and delayed recall when they studied by review and when they studied by taking a free recall test.
A sample of 90 undergraduate students read and studied two passages. They studied one passage by repeated reading, and the other by repeated testing. After studying, they took two short answer comprehension tests on each passage, the first immediately after reading the passage and the second after a week long delay. The independent variables included time of test, method of study, type of question (factual recall or inference), language comprehension, background knowledge, and participants' metacognitive judgments about their learning and future performance. The dependent variable was performance on the comprehension tests.
Results indicated that a testing effect was not clearly established. Although there was less forgetting in the repeated test condition than in the repeated review condition, which is consistent with a testing effect, the review condition consistently outperformed the testing condition, which is not consistent with the testing effect. Differences between the results of this study and those from other testing effect research are hypothesized to be due to the detailed nature of the test questions. Regarding the cognitive variables, metacognitive judgments of learning (JOLs) were not consistently predictive of test performance, background knowledge predicted performance on tests in the repeated review but not the repeated test conditions, and language comprehension, which was the best predictor overall, was a significant predictor of performance on the immediate test in the repeated test condition, and the delayed tests in both conditions. Future research should focus on examining the effects of individual differences on the efficacy of studying via retrieval on immediate and delayed recall.


  • thumnail for Kern_columbia_0054D_12380.pdf Kern_columbia_0054D_12380.pdf application/pdf 2.37 MB Download File

More About This Work

Academic Units
School Psychology
Thesis Advisors
Peverly, Stephen T.
Ph.D., Columbia University
Published Here
October 13, 2014