For our February 20th meeting, we will be reading:
Zamary, A. & Rawson, K.A. (2016). Which technique is most effective for learning declarative concepts – Provided examples, generated examples, or both? Educational Psychology Review, 30(1), 275-301. doi: 10.1007/s10648-016-9396-9.
In this study, researchers examined whether studying by generating examples of concepts would be better for learning than studying provided examples of concepts. They tested this by assigning participants to study by (1) reviewing four provided examples of 10 concepts from a social psychology textbook chapter, or (2) generating four of their own examples of each concept, or by combining a provided example with a generated example for each concept. They hypothesized that generating examples would lead to better learning than studying provided examples. They further supposed that perhaps using both study methods may have an additive benefit on learning. Participants were asked to categorize examples of the concepts and provide definitions for the concepts two days after study. The results indicated that studying using provided examples resulted in higher scores on the categorization test than either of the other study methods. In addition, studying using provided examples took the least amount of time. In study 2, the researchers added a fourth condition – simultaneously viewing a provided example and generating an example. This fourth condition was also less effective and less efficient than studying provided examples.
- Exciting to see research exploring how students best learn the foundations for classes. Much research focuses on “active learning” or “higher-order processing” rather than how students acquire the basic knowledge they need to engage in more critical thinking.
- The researchers devised a measure of efficiency in terms of amount learned per unit time (“gain per minute”). This focus on efficiency – an element that can help instructors persuade their students to use better study habits – is often not accounted for in SoTL studies.
- We noted a number of flaws in the statistical analyses. In using a directional test, it is inappropriate to make conclusions about a result in the opposite of the prediction direction. The researchers also appear to have run an ANOVA first to test for order effects, but then persist with their directional t-test analyses. In addition, the pattern of p-values for some effects (in the range of .02-.05) gave fairly weak evidence. Lastly, to improve the strength of the research, Study 2 could have been pre-registered and a proper power analysis conducted prior to data collection to ensure an adequate sample size.
- One potential confound is that the test (categorization of examples) was more similar to the provided examples study method than the generated example study method. In other words, the results could have been due to the fact that the study-test conditions matched for provided examples but not for generated ones. The recall data are consistent with the idea that the categorization effects were confounded. If the provided examples really produced better learning, we might have expected to them also lead to better cued recall.
Take Home Message:
Providing students with examples of concepts requires (a lot!) less study time and produces learning at least as good as making the students generate their own examples. Future research should examine the “dosage” effect of provided examples to help teachers optimize efficiency.