Date of Award


Degree Type


Degree Name

Master of Science (MS)


Psychological Sciences

Committee Chair

Jeffrey D. Karpicke

Committee Member 1

Thomas S. Redick

Committee Member 2

James S. Nairne


Recently, van Gog and Sweller (2015) asserted that the advantages of testing diminish (or even disappear altogether) as the complexity of learning materials increases. To elucidate their claim, they used the term “element interactivity” as a proxy for material complexity. While material low in element interactivity can be thought of as a series of isolated facts, material that is high in element interactivity contains strongly related ideas such that the learning of any particular idea is contingent on understanding other components of the material. The current set of experiments systematically manipulated element interactivity in order to evaluate the validity of van Gog and Sweller’s contention. Experiment 1 manipulated element interactivity by scrambling the order of sentences within educational texts. Specifically, students studied two educational texts that were either presented intact or with their sentences scrambled. For one of the passages, students engaged in retrieval practice following study, and for the other, they completed a distractor task following study. Subjects’ memory for the passage content was assessed a week later when they were asked to answer a series of questions about the texts and freely recall the information they studied the previous week. Although it may seem counterintuitive, the logic put forth by van Gog and Sweller (2015) would argue that for the scrambled passages (which were lower in element interactivity) testing effects would be present, and for the passages presented with their sentences intact (higher element interactivity), testing effects should be absent. Contrary to this prediction, doing retrieval practice led to enhanced performance on all sections of the final test regardless of whether the texts were intact or scrambled. Experiment 2 manipulated element interactivity by altering subjects’ contextual prior knowledge. Participants studied an ambiguously worded passage in the presence or absence of a topic word that provided a relational schema to guide their interpretation. Because the topic word served to create relations among the ideas presented in the text, van Gog and Sweller’s hypothesis would assert that retrieval practice effects should be absent when the topic word is visible and present when it is not visible. As in Experiment 1, subjects either did retrieval practice or completed a distractor task after studying the passage and took a final test roughly one week later. Results showed a benefit of retrieval practice for both the topic absent and the topic present groups. Experiment 3 was largely identical to Experiment 2 in that it manipulated subjects’ contextual prior knowledge and followed a similar procedure. However, in this case, participants studied word lists that conformed to ad hoc categories in the presence or absence of the category names. In terms of predictions, we speculated that giving subjects’ access to the category names would increase element interactivity, thereby implying that van Gog and Sweller (2015) would argue that testing effects should be absent when category names are present. In stark contrast to this assertion, findings from this experiment indicated a benefit of engaging in retrieval practice regardless of whether the category names were present or not. Nevertheless, there were no differences in performance between the category names present and category names absent groups. Thus, across three experiments, no evidence was found to support the contention that retrieval practice effects are absent when using complex materials.