Improving encoding strategies as a function of test knowledge and experience

Abstract

Information that is produced or generated during learning is better remembered than information that is passively read, a phenomenon known as the generation effect. Prior research by deWinstanley and Bjork (Memory & Cognition, 32, 945-955, 2004) has shown that learners, after experiencing the memorial benefits of generation in the context of a fillin-the-blank test following the study of a text passage containing both to-be-read and to-be-generated items, become more effective encoders of to-be-read items on a second passage, thus eliminating the generation effect on a subsequent memory test. Current explanations of this phenomenon assume that learners need to actually experience the generation advantage on the test of the first passage to become more effective encoders of to-be-read items on the second passage. The results of the present research, however, suggest otherwise. Although experiencing a test of the first passage does appear to be critical for leading participants to become better encoders on the second passage, experiencing a generation advantage on the test for the first passage is not. More generally, these results shine new light on the generation effect as well as how and why taking tests has the potential to improve subsequent learning.

Lecturer fluency can impact students' judgments of learning and actual learning performance

“…The main effects of time on JOLs and actual learning performance are consistent with previous research indicating that the act of taking a test can reduce overconfidence (Yue et al, 2015) and improve encoding (Storm et al, 2016)—in this case, these effects occurred even when the material was delivered by a fluent or disfluent lecturer and when the content of the two tests (and videos) were unrelated.…”

“…However, this hypothesis was rendered moot as those in the fluent and disfluent conditions were equally overconfident—it is difficult to remedy an asymmetry that does not exist. That said, JOLs were lower for the second video across all three experiments indicating that the experience of taking a test can reduce overconfidence (Storm et al, 2016), even when the repeated tests cover different material. Further, in Experiment 2, we found that participants improved calibration for the second video, especially after a delay.…”

“…We were also interested in determining whether providing student‐participants with two opportunities to make a JOL over different lecture videos with an intervening test would allow them to address overconfidence stemming from the fluent lecturer (even without feedback). We hypothesized that overconfidence could be addressed via a combination of participants improving their encoding strategies after a test (Storm et al, 2016) and lowering their JOLs (Yue et al, 2015). Thus, in our studies, participants were provided with two opportunities to watch a video and make a related JOL.…”

“…While no research to‐date has examined repeated testing as a potential “treatment” for the overconfidence caused by fluent lecturers, other studies have examined the impact of test‐taking on learning performance more generally. Previous research has indicated that the act of taking a test can result in improved memory performance even without feedback; essentially, test‐taking can help people improve their encoding strategies for future learning episodes (Storm, Hickman, & Bjork, 2016). This advantage can be observed when learners are made aware of the test format, and consequently, the type of information to which the test is sensitive.…”

Test expectancy and memory for important information.

“…Namely, participants were less selective in their recall when their previous testing experience consisted of recognition tests than of recall tests. This would suggest that participants learned from testing and adjusted their strategies accordingly (Garcia-Marques et al, 2015; Jensen, McDaniel, Woodard, & Kummer, 2014; Storm et al, 2016). That RgRc participants were not as selective as the All Rc participants indicates that they did not learn to prioritize high-value items simply as a consequence of study design—such prioritization was unnecessary when their test performance was at ceiling—and thus did not have the same amount of practice executing a value-based strategy as those in the All Rc condition, for whom such a strategy was always important.…”