A common belief amongst students and teachers is that testing serves primarily as an evaluative tool, as a “dipstick” to assess what one knows and does not know (Hartwig & Dunlosky, 2012; McAndrew, Morrow, Atiyeh, & Pierre, 2016; Morehead, Rhodes, & DeLozier, 2016). However, there is a voluminous literature demonstrating, both in the laboratory and in the classroom, that taking an initial test over previously encountered material actually potentiates learning compared to simply restudying the material. This common and robust finding has been referred to as the testing effect, or retrieval practice (for recent reviews, see Carpenter, 2012; Dunlosky, Rawson, Marsh, Nathan, & Willingham, 2013; Karpicke, in press; Kornell & Vaughn, 2016; McDermott, Arnold, & Nelson, 2014; Roediger, Putnam, & Smith, 2011; Rowland, 2014).
Classroom instruction could benefit greatly from this easy to implement and inexpensive technique (Roediger & Pyc, 2012). Science, technology, engineering, and mathematics (STEM) classrooms, in particular, are in need of instructional approaches that can promote student achievement, as students struggle to gain proficiency and have increased anxiety over the subject matter (Ackerman, Kanfer, & Beier, 2013; Chen, 2013). Toward this goal, in recent studies retrieval practice has been shown to potentiate student learning in courses such as biology, biochemistry, chemistry, and engineering (Butler, Marsh, Slavinsky, & Baraniuk, 2014; Carpenter et al., 2016; Horn & Hernick, 2015; Pyburn, Pazicni, Benassi, & Tappin, 2014).
While the retrospective benefits of testing have been clearly demonstrated, an open question concerns the prospective benefits of testing. That is, does the positive effect of testing also extend to testing that takes place before exposure to material that must be learned? There is evidence that giving tests prior to learning improves memory for that information. In a typical study using these “prequestions”, individuals read passages of text and then complete a test over the material. Half the participants receive prequestions before each to-be-read segment (i.e., the prequestion group), and the other half read each segment without receiving pre-questions (i.e., the control group). On the final test, the prequestion group typically outperforms the control group (Little & Bjork, 2016; Peeck, 1970; Pressley, Tanenbaum, McDaniel, & Wood, 1990; Richland, Kornell, & Kao, 2009; Rickards, Anderson, & McCormick, 1976).
In these studies, corrective feedback was not provided to students at the time they answered the prequestions. Instead, to learn the information students had to discover it while they read the material. Even though participants often got the prequestion wrong—because they had not yet learned the information—the better learning that occurred in the prequestion group relative to the control group demonstrates that students can successfully discover information needed to answer the prequestion, and can retain this information at a later time. Students who receive prequestions typically show better memory for portions of the passage that were relevant to the prequestions—i.e., prequestioned information—compared to other portions of the passage that were not relevant—i.e., non-prequestioned information (e.g., Bull & Dizney, 1973; Pressley et al., 1990; Richland et al., 2009). In addition, some research shows that under some circumstances the prequestion group retains non-prequestioned information better than the control group (Carpenter & Toftness, 2017), suggesting that prequestions may serve to enhance overall processing of the material.
Given its simplicity and apparent effectiveness, the prequestion technique would seem to be a useful tool that can be implemented to enhance student learning in instructional settings. Indeed, a Department of Education-sponsored practice guide for educators lists prequestions as one of a handful of evidence-based techniques, judged by a panel of experts on learning, that are concrete and highly applicable to education (Pashler et al., 2007). Though the benefits of prequestions have been consistently demonstrated in laboratory studies with reading materials, the expert panel noted that the level of ecologically valid evidence supporting the prequestion technique has been low. In particular, they concluded that “there is little or no published experimental evidence regarding whether pre-questions will promote the learning of orally presented classroom content,” (p. 19).
Indeed, in the 10 years since the publication of the practice guide, the potential of prequestions to enhance learning in educational settings has remained largely unexplored. To date, only one known study has explored prequestions in a classroom setting. McDaniel, Agarwal, Huelser, McDermott, and Roediger (2011) (experiments 2a and 2b) presented middle school science students with questions at the start of each lesson and end of each lesson (prelesson-postlesson questions), or only at the end of each lesson (postlesson questions). Across two experiments, they found mixed evidence for a benefit of prequestions. On the postlesson questions, performance was generally higher for material that had been prequestioned (i.e., prelesson-postlesson questions) compared to material that had not been prequestioned (i.e., postlesson questions). There was a numeric, but non-significant, advantage in experiment 2a (78 vs. 76%, respectively, d = 0.15), and a modest but significant advantage in experiment 2b (84 vs. 79%, respectively, d = 0.35). Thus, contrary to the results of laboratory studies where the benefits of prequestioned information over non-prequestioned information have been quite large—e.g., 25% in the study by Richland et al. (2009) (d = 1.70)—these effects appear to be much smaller in classroom settings.
Such a reduction in the size of the effect between laboratory and classroom settings is not particularly surprising given a multitude of factors that can be controlled in the laboratory but not in the classroom (e.g., see Butler et al., 2014). Potential sources of unexplained variance in course settings could include students’ prior knowledge of the material, out-of-class studying, and individual differences in interest, motivation, and academic achievement.
Beyond these factors, differences in the way that prequestions are administered could lead to different effects on learning as well. In particular, the benefits of prequestions could be reduced under conditions in which the mechanism(s) believed to be responsible for these effects are less likely to be engaged. Theoretical discussions of the prequestion effect have included the (non-mutually exclusive) possibilities that prequestions stimulate curiosity (e.g., Berlyne, 1954, 1962), that they increase attention to those parts of the material that are relevant to the prequestions (Peeck, 1970; Pressley et al., 1990), or that they provide a metacognitive “reality check” (e.g., Bjork, Dunlosky, & Kornell, 2013) that gives students a clear realization that they do not know the answers and must allocate attention and effort to discover them. Such processes are most likely to occur under conditions in which students’ curiosity to know the answers to the prequestions is high, they are attentive to the material, and they can successfully discover the answers to the prequestions during the learning episode.
These things considered, it is quite possible that the effects of prequestions are consistently limited in classroom settings. Compared to a laboratory setting, a classroom is likely to involve presentation of information that is more lengthy and complex. During a 50-minute (or longer) class period, students may have difficulty sustaining attention and noticing information that is relevant to a prequestion that was asked at the beginning of class. Even if curiosity to know the answer is high, this rather lengthy duration of time may introduce interruptions in the processing of the prequestion and the ease with which students can remember it and connect it to the relevant information in the lesson.
In the current study, we set out to provide additional data on the effectiveness of prequestions in classroom settings. In a college-level course on chemical engineering, students were asked a question at the beginning of several class meetings that pertained to that day’s lesson. At the end of class, they were asked the same question again, in addition to a different question from the same lesson. Consistent with laboratory studies on prequestions, we did not provide students with feedback of the correct answers at the time of the prequestions. The answer to a given prequestion was always contained in the class lesson that immediately followed, and to learn the answer students needed to discover it during class. This was somewhat different from the study by McDaniel et al. (2011), in which feedback was provided after the prequestions and the lessons relevant to some of the prequestions occurred during subsequent class meetings (these were necessary design features of the study that aligned with the way in which the course was structured). However, common to the current study and that of McDaniel et al. (2011) was the fact that prequestions were administered in real classrooms where the duration of a lesson was longer than the duration of a typical laboratory experiment. If the length and complexity of information presented in authentic course environments is a contributing factor to the attenuation of prequestion effects, then the current study might be expected to yield results similar to those of McDaniel et al. (2011). To the extent that feedback is important to these effects—such that withholding feedback produces benefits of prequestions, perhaps due to enhanced curiosity (e.g., Berlyne, 1954, 1962)—then the current study might be more likely to yield benefits of prequestions.
Given the variability among students in classroom environments, we also explored the potential role of individual differences in the prequestion effect. We collected information about students’ academic achievement (i.e., grade point average (GPA)), in addition to their confidence in their answers to the prequestions, familiarity with the information in the prequestions, and how much of the assigned reading they had completed prior to class. Such individual differences have not been examined in any of the known research on prequestions, but could be of theoretical and practical importance. In particular, the idea that prequestions provide a metacognitive reality check that reduces students’ overconfidence could be tested by examining the relationship between students’ confidence in their answers to the prequestions and their later accuracy on those questions at the end of class. A negative relationship between confidence and later accuracy (other factors controlled) might offer some support for this notion, in that students who are less certain about their knowledge of the concept in the prequestion are more likely to learn and retain the answer to that question when it is encountered in class.
Abundant research on retrieval practice shows that testing students over material they have encountered boosts memory retention (e.g., Carpenter, 2012; Dunlosky et al., 2013; Karpicke, in press; Kornell & Vaughn, 2016; McDermott et al., 2014; Roediger, Agarwal, McDaniel, & McDermott, 2011; Rowland, 2014). However, it is unknown whether these benefits are enhanced by giving students a chance to answer the test questions (without feedback) before encountering the material during class. To examine this, we gave students a practice quiz at the end of each week of the course. The quiz included the questions that were asked at the beginning and end of class, along with questions that were asked only at the end of class, and never-before-seen questions that were drawn from the same lessons. This provided the opportunity to examine delayed retention for material that was tested vs. material that was not tested—i.e., retrieval practice—and to determine whether this effect was stronger for material that was prequestioned before it was tested at the end of class.
Thus, the current study provided a classroom investigation of the effects of prequestions on learning course information. In addition, we examined whether or not these effects are influenced by individual differences in students’ confidence, academic achievement, and out-of-class preparation. Finally, we examined whether the common benefits of retrieval practice—superior long-term retention for information that has been tested—could be enhanced by providing students with prequestions prior to the lesson.