Four semesters investigating frequency of testing, the testing effect, and transfer of training.

Abstract

We carried out 4 semester-long studies of student performance in a college research methods course (total N ϭ 588). Two sections of it were taught each semester with systematic and controlled differences between them. Key manipulations were repeated (with some variation) across the 4 terms, allowing assessment of replicability of effects. Variables studied included frequency of tests (e.g., 2 vs. 8 in-class exams), the repetition of some and not other exam items (i.e., the testing effect), and variation of test items between the in-class exams and the final exam (e.g., identical items vs. controlled changes in items). Some studies also manipulated presence or absence of low-stakes quizzes. The repetition of test items generally led to better performance. However, we did not observe consistent superiority for items that were repeated exactly over those that were repeated in modified form; the reverse was more often the case. The effect of the low-stakes quizzes was minimal at best. Results are discussed in terms of memory and transfer of training models. Educational Impact and Implications StatementCan we find inexpensive and easily adaptable modifications to teaching methods that positively impact student outcomes? These studies provide a positive answer to that question. The work is based on laboratory findings that frequent tests and frequent attempts to recall the same material (1) aid learning and memory, and (2) help students apply what they've learned to new problems. The present studies took place in large-enrollment college classes across four semesters. Within each semester two sections of an undergraduate course were taught in a highly similar fashion, primarily differing in the number of tests given and whether items that appeared on an earlier test were repeated on the final exam. In addition, some of the repeated items were identically so, while other 'repeated' items tested the same concepts but with different wording. We found evidence that frequent testing and repetition of tested items can improve course performance up to about 10%, though the results varied across the studies so further work is needed to clarify why. We also observed that under some circumstances students did as well or even better on re-worded test items as they did when the item was repeated in exactly the same words.

Putting evidence‐based claims to the test: A multi‐site classroom study of retrieval practice and spaced practice

“…Although the lack of overlap information should be rectified in future work, there is evidence that retrieval practice can make a difference even when the test questions are not the same as the quiz questions (i.e., when they are new, related questions from the textbook, Batsell, Perry, Hanley, & Hostetter, ). In fact, Foss and Pirozzolo () did not observe consistent superiority for items that were repeated exactly over those that were repeated in modified form; the reverse was more often the case.…”

Frequent mastery testing with second‐chance exams leads to enhanced student learning in undergraduate engineering

“…A meta-analysis by Bangert-Drowns, Kulik, and Kulik (1991) found large learning gains for testing over no testing, but that increasing the frequency of assessment beyond three assessments resulted in smaller learning gains. More recently, Foss and Pirozzolo (2017), found that across four semesters students assessed more frequently scored higher on a final exam with a small effect size (d = 0.16). However, the difference between the conditions was only significant for one of the semesters.…”

Examining the Testing Effect in University Teaching: Retrievability and Question Format Matter

“…Other studies are limited because they lack a restudy control condition but compare the testing condition to conditions in which no exposure to information subsequent to the initial learning took place (McDaniel et al, 2007a, 2011, 2013; Johnson and Kiviniemi, 2009; Mayer et al, 2009; Vojdanoska et al, 2010; Lyle and Crawford, 2011; Roediger et al, 2011; Marsh et al, 2012; Shapiro and Gordon, 2012; Bell et al, 2015; Downs, 2015; Khanna, 2015; Batsell et al, 2017; Foss and Pirozzolo, 2017). In these studies, the testing effect is confounded with differences in exposure to and engagement with learning content, which severely limits the interpretation of their findings.…”