Role of expectations and explanations in learning by teaching

Contemporary Educational Psychology 39 (2014) 75–85

Contents lists available at ScienceDirect

Contemporary Educational Psychology journal homepage: www.elsevier.com/locate/cedpsych

Role of expectations and explanations in learning by teaching Logan Fiorella ⇑, Richard E. Mayer University of California, Santa Barbara, United States

a r t i c l e

i n f o

Article history: Available online 18 January 2014 Keywords: Learning by teaching Explaining Preparing to teach Learning strategies

a b s t r a c t The present study examined the role of preparing to teach (i.e., teaching expectancy) and actually teaching (i.e., explaining to others) on immediate and long-term learning. In Experiment 1, participants studied a base version or an enhanced version of a paper-based lesson on how the Doppler Effect works with the expectation of taking a test on the material or with the expectation of teaching the material by providing a video-recorded lecture. Results indicated that those who prepared to teach (without actually teaching) outperformed those who prepared for a test on an immediate comprehension test (i.e., a teaching expectancy effect; d = .55), regardless of the format of the lesson. In Experiment 2, participants studied while expecting to be tested or expecting to teach the material; some then actually did teach the material by providing a video-recorded lecture, whereas others received additional study time. Results indicated that those who actually taught the material outperformed those who did not teach on a delayed comprehension test (i.e., a teaching effect; d = .56), though this effect was strongest for those who also prepared to teach. Overall, these findings are consistent with the idea that preparing to teach results in short-term learning gains, whereas the act of teaching (i.e., by explaining the material to others) coupled with preparing to teach is important for long-term learning. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction It is often said that the best way to learn something new is to teach it to someone else. Thus, it is not surprising that learning by teaching is prevalent among contemporary educational practices, including peer tutoring (e.g., Palincsar & Brown, 1984), cooperative learning environments (e.g., Slavin, 1983), and even interacting with computer-based pedagogical agents (e.g., Biswas, Leelawong, Schwartz, Vye, & The Teachable Agents Group at Vanderbilt, 2005). Although there is considerable evidence that teaching can enhance one’s own learning (e.g., Allen & Feldman, 1973; Cohen, Kulik, & Kulik, 1982; Gartner, Kohler, & Riessman, 1971; King, Staffieri, & Adelgais, 1998; Morgan & Toy, 1970; Robinson, Schoefield, & Steers-Wentzell, 2005; Rohrbeck, Ginsburg-Block, Fantuzzo, & Miller, 2003; Roscoe & Chi, 2007), it is largely unclear how different components of the teaching process influence learning (Galbraith & Winterbottom, 2011; Rohrbeck et al., 2003; Roscoe & Chi, 2007). This ambiguity is likely due to the diversity of learning by teaching programs available (e.g., cross-age tutoring, reciprocal tutoring, teachable agents), the many teaching-related activities

⇑ Corresponding author. Address: Department of Psychological and Brain Sciences, University of California, Santa Barbara, CA 93106, United States. E-mail addresses: fi[email protected] (L. Fiorella), [email protected] (R.E. Mayer). http://dx.doi.org/10.1016/j.cedpsych.2014.01.001 0361-476X/Ó 2014 Elsevier Inc. All rights reserved.

potentially responsible for learning (e.g., preparing to teach, explaining to others, providing feedback), as well as other interactions between students that may or may not be unique to teaching (e.g., answering questions), but that still may contribute to learning. In addition, much of the evidence for learning by teaching is derived from research evaluating the effectiveness of peer tutoring programs over conventional instruction (Roscoe & Chi, 2007). Although this program-based approach may offer important practical implications for implementing specific learning by teaching programs into the classroom, it is limited in its ability to explain how distinct features of learning by teaching differentially influence student learning (such as preparing to teach versus explaining to others). An alternative approach is to break down the teaching process into distinct stages and to systematically test how each stage uniquely contributes to learning. For example, Bargh and Schul (1980) offered an early framework that proposed three distinct stages of teaching: preparing to teach, explaining to others, and interacting with others. First, preparing to teach (without actually teaching) may improve learning because the expectation of teaching encourages learners to better select and organize to-be-learned material during learning. Second, explaining the material to others (without interacting with others) may offer additional benefits because it encourages learners to elaborate on the material presented and to make connections between the to-be-learned material and their existing knowledge. Finally, interacting with others—by

76

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85

answering questions or providing feedback—may offer metacognitive benefits by encouraging learners to reflect on their own understanding of the material and to identify gaps in their knowledge. Yet relatively little research has followed this stage-based approach (e.g., Annis, 1983; Fiorella & Mayer, 2013; Coleman, Brown, & Rivkin, 1997; Roscoe & Chi, 2008), and as a result, it is largely unclear the extent to which effects of learning by teaching can be attributed to the mere expectation of teaching (i.e., preparing to teach), the act of teaching (i.e., explaining to others), or interactions between students that may or may not be unique to teaching (e.g., answering questions and receiving feedback). The current study focused on contributing toward a better understanding of the first two stages of learning by teaching—preparing to teach and explaining to others. In Experiment 1, participants studied two versions of a paper-based lesson on how the Doppler Effect works with instructions that they would later be tested on the material or asked to teach the material by providing a short video-recorded lecture. Those expecting to teach did not actually teach, but instead all participants were given an immediate comprehension test on the material. In Experiment 2, participants were given the same instructions (i.e., prepare for a test or prepare to teach) but some actually did teach the material by providing a video-recorded lecture, whereas others only restudied the material. All participants were then assessed on a delayed comprehension test of the material. Overall, the goal of the current study was to examine the effects of preparing to teach and the effects of explaining to others on immediate and long-term learning. 1.1. Learning by preparing to teach An often-overlooked factor in learning by teaching is the unique effect of preparing to teach on learning. Indeed, research on the teaching expectancy effect suggests that studying with the mere expectation of later teaching can enhance learning beyond studying normally for a test (e.g., Fiorella & Mayer, 2013; Bargh & Schul, 1980; Benware & Deci, 1984). For example, in a classic study by Bargh and Schul (1980), students were given verbal material to study with the expectation of either answering questions afterwards or teaching the material to another student. Those who prepared to teach outperformed those who prepared to be tested on subsequent recall and recognition tests. Benware and Deci (1984) replicated this finding with more meaningful learning materials when they asked students to study an article on brain functioning with the expectation of teaching or being tested on the material. Those who expected to teach performed better on a conceptual knowledge test than those who expected to be tested. These studies indicate that preparing to teach plays an important role in learning by teaching. At the same time, preparing to teach may not result in improved learning under some conditions (Fiorella & Mayer, 2013; Ehly, Keith, & Bratton, 1987; Renkl, 1995). For example, in a study by Renkl (1995), students prepared to teach others about probability while studying worked-out examples on a computer. The results showed no evidence for a teaching expectancy effect, likely due to the fact that students may have experienced excessive stress and anxiety at the prospect of teaching others. Further, a recent study by Fiorella and Mayer (2013) suggests that the teaching expectancy effect may also be limited to short-term learning benefits. In their study, participants studied a lesson on the Doppler Effect with the expectation of later teaching or being tested on the material. Those who prepared to teach (without actually teaching) significantly outperformed those who prepared to be tested on an immediate comprehension test (d = .59); however, this effect did not reach statistical significance when students were assessed following a 1-week delay (d = .24). These findings are also similar to those reported by Annis (1983), which indicated only mixed evi-

dence for a teaching expectancy effect when learning was assessed after a delay. 1.2. Learning by explaining to others Explaining material to others may offer unique learning benefits beyond only preparing to teach (Fiorella & Mayer, 2013; Bargh & Schul, 1980). This is because generating explanations requires learners to go beyond the information presented and make connections between to-be-learned material and their existing knowledge (Chi, 2000). In other words, learning by explaining depends on the extent to which learners reflect on their own understanding and integrate the material with their prior knowledge—what is sometimes referred to as reflective knowledge building (Roscoe & Chi, 2007). On the other hand, learners are unlikely to benefit from engaging in knowledge telling—that is, simply summarizing the material and making only minor inferences. In short, generating quality explanations for others is likely to result in deep learning. Although there is much research on the benefits of self-explanation (Fonseca & Chi, 2011; Roy & Chi, 2005), relatively few studies have systematically investigated the effects of explaining to others (Annis, 1983; Coleman, Brown, & Rivkin, 1997; Roscoe & Chi, 2008). Further, one important limitation of this research is that the effects of explaining to others are often not isolated from potential effects of other stages of learning by teaching (e.g., preparing to teach or interacting with others). For example, in the study by Annis (1983), participants studied a history lesson either with the expectation of later teaching another student or with the expectation of being tested on the material. Of those expecting to teach, some actually did teach another student, whereas others only prepared to teach. Results indicated that students who taught generally outperformed those who only prepared to teach on delayed measures of learning. However, the students who taught did so by interacting with a peer who was encouraged to ask questions throughout the teaching session. According to Bargh and Schul’s (1980) framework, interacting with others represents a distinct stage of the teaching process; thus, it is unclear whether the added benefit of teaching was due to explaining the material to another person, interacting with another person, or some combination between the two. In a study by Coleman, Brown, and Rivkin (1997), students learned about evolution by natural selection and then either selfexplained the material or explained the material to another student (without interacting with the other student). Results indicated that students who explained to a peer generally outperformed students who self-explained on measures of deep learning. However, one limitation of this study is that the groups studied the material with different expectations—that is, the selfexplain group studied with the expectation of later self-explaining, whereas the explain-to-other group studied with the expectation that they would explain the material to another student. Thus, it is unclear the extent to which the potential benefits of preparing to teach may have influenced the results. One approach to isolate the effects of explaining to others is to not inform participants that they will later teach the material. For example, in a study by Roscoe and Chi (2008), students learned about the human circulatory system by studying a lesson normally. They were then asked to explain the material to themselves, explain the material by giving a video-recorded lecture, or explain the material to another student while also interacting with the student (e.g., by answering questions). Results indicated that those who self-explained generally outperformed those who explained to others on measures of deep learning. Self-explainers also showed more evidence of reflective knowledge building than the explain-to-other groups. However, it may be that the benefits of explaining to others depend on also having studied the material with the expectation of later teaching.

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85

Finally, the study by Fiorella and Mayer (2013) indicated that preparing to teach and actually teaching had similar benefits when learning was assessed on an immediate comprehension test; however, only those who both prepared to teach and actually taught the material experienced learning benefits after a 1-week delay. This finding suggests that one important added benefit of explaining material to others is that it promotes long-term learning. 1.3. Theory and predictions According to the cognitive theory of multimedia learning (Mayer, 2005, 2009, 2011), meaningful learning depends on learners’ ability to select the most relevant information from a lesson, organize it into a meaningful cognitive representation, and integrate it with their prior knowledge. Further, the cognitive theory of multimedia learning—similar to cognitive load theory (Sweller, 2005; Sweller, Ayres, & Kalyuga, 2011)—is based on the idea that learners have a very limited processing capacity, which they must use to engage in cognitive processing necessary for learning. Thus, the goal of learning strategies such as learning by teaching is to encourage learners to invest any remaining cognitive capacity after initially representing the material (i.e., corresponding to the cognitive process of selecting) to deeper processing directed toward making sense out of the material (i.e., corresponding to the cognitive processes of organizing and integrating)—which is referred to as generative processing. Although generative processing can be difficult to measure directly and reliably, it can be reasonably inferred from performance on measures of meaningful learning outcomes, such as long-term comprehension or transfer. Such assessments are intended to target understanding of important concepts rather than rote memorization of basic facts. The current study aimed to better understand the extent to which preparing to teach and actually teaching (i.e., explaining to others) foster generative processing, as indicated by performance on a comprehension test. Experiment 1 extends previous research by examining the robustness of the teaching expectancy effect across standard and enhanced versions of a paper-based lesson on the Doppler Effect, in which the enhanced version contained more support for organizing the lesson. In a 2  2 design, participants studied either a base version or an enhanced version of a paper-based lesson on how the Doppler Effect works with the expectation of later teaching (i.e., by providing a video-recorded lecture) or being tested on the material; all students then took an immediate comprehension test on the material. The base version of the lesson was intended to require learners to invest more effort toward organizing the to-belearned material by providing the essential information in absence of any instructional support. The enhanced version of the lesson was intended to require less cognitive demands to organize the to-be-learned material. In particular, the enhanced lesson included signaling (i.e., section headings and numbering) designed to direct learners’ attention toward the organizational features of the lesson, the lesson was segmented into smaller, more manageable sections (rather than being presented as one long section), and the lesson included graphics depicting the material described in the text to help learners construct a coherent representation of the material. We predicted that the teaching expectancy effect may be strongest when students learned from a base lesson, because the act of preparing encouraged them to better organize the material, as originally proposed by Bargh and Schul (1980). Experiment 2 extends previous research by testing whether the long-term benefits of actually teaching depend on whether students also prepared to teach during learning. In a 2  2 design, participants studied a lesson on the Doppler Effect with the expectation of teaching or being tested on the material; students then either taught the material by providing a video-recorded lecture, or they did not teach the material and instead received addi-

77

tional study time. If the benefits of teaching depend on preparing to teach, the teaching effect will be strongest for students who both prepared to teach and actually taught. Alternatively, if the act of teaching is the primary driver of long-term learning, the teaching effect will take place regardless of whether students prepared to teach or studied the material normally for a test. Further, those who only prepared to teach without actually teaching will not show benefits beyond studying normally on a delayed test. Overall, the current study sought to better understand the relationship between preparing to teach and explaining to others on immediate and long-term learning. 2. Experiment 1 The primary goal of Experiment 1 was to determine whether the teaching expectancy effect is strongest when students study the base version of the lesson on the Doppler Effect compared to when students study the enhanced version of the lesson. 2.1. Method 2.1.1. Participants and design The participants were 95 undergraduate students recruited from the Psychology Subject Pool at a western university. The mean age was 19.1 years (SD = 1.4), and there were 41 men and 54 women. The groups did not differ significantly in terms of mean age or proportion of men and women. The study followed a 2  2 design, in which expectation (i.e., teach or test) and lesson format (i.e., base version or enhanced version) served as between-subjects factors. Forty-eight participants expected to teach the material following studying (expect teach group), whereas 47 expected to be tested on the material (expect test group). Of those in the expect teach group, 24 studied a base version of the lesson (expect teach—base group), and 24 studied an enhanced version of the lesson (expect teach—enhanced group). Of those in the expect test group, 23 studied a base version of the lesson (expect test—base group), and 24 studied an enhanced version of the lesson (expect test—enhanced group). Participants had relatively low prior knowledge of the Doppler Effect, as reported on a questionnaire with a possible score of 13 (expect teach—base: M = 5.3, SD = 2.8; expect teach—enhanced: M = 5.2, SD = 2.7; expect test—base: M = 6.3, SD = 2.6; expect test—enhanced: M = 5.7, SD = 2.6). The groups did not significantly differ on level of prior knowledge. 2.1.2. Materials The paper-based materials consisted of a consent form, a demographics form, an instructions sheet, two versions of a paper-based lesson on the Doppler Effect, a comprehension test, and a postquestionnaire. The consent form described the details of the study, informed participants they may be videotaped during the experiment and that their privacy was protected, and included a place for them to sign. The demographics questionnaire asked participants to provide their age, gender, and SAT scores. They were also asked to rate their knowledge of the Doppler Effect on a 5-point scale from 1 (‘‘very low’’) to 5 (‘‘very high’’) and to place a check mark next to each of the following items that applied to them: ‘‘I have taken a college course in physics,’’ ‘‘I know what Hz means,’’ ‘‘I have used an oscilloscope,’’ ‘‘I know how radar works,’’ ‘‘I know the basic characteristics of sound waves,’’ ‘‘I know what relative motion is,’’ ‘‘I know what the red shift is,’’ and ‘‘I know what a sine curve is.’’ The self-reported rating of prior knowledge and the checklist served as a measure of participants’ prior knowledge of the Doppler Effect, with a combined possible score of 13 (i.e., one point for each checked item on the list plus 1 through 5 based on which level was checked on the rating scale with 5 = very high

78

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85

and 1 = very low). This more indirect measure of prior knowledge was used to avoid potential testing effects, in which taking a practice test can serve as a form of instruction (Roediger & Karpicke, 2006). The test also correlated highly with performance on the final comprehension test in both experiments (Exp. 1: r = .78, p < .05; Exp. 2: r = .69). Finally, the demographics form asked participants to report their self-efficacy by rating how well they thought they would perform on a test after studying a lesson on the Doppler Effect on a 5-point scale from 1 (very poorly) to 5 (very well). This item was used to determine if the groups were equated on reported self-efficacy. The instructions sheet described the task details of the experiment. Participants received one of two versions depending on which condition they were randomly assigned. The expect test group was told that they would be expected to answer questions on the material they studied: In this experiment, you will study a short lesson on how the Doppler Effect works and then be asked to answer some questions about what you learned. You will have 7 minutes to study the lesson. You may take notes on the lesson itself and/or use the blank sheet of paper provided to you by the experimenter; however, you will not be able to use your notes while answering questions on the material. Please continue working until your time has expired. If you have any questions, please ask the experimenter now. He/she will not be able to answer questions once you begin. Participants assigned to the expect teach group were told that they would be expected to teach the material they studied: In this experiment, you will study a short lesson on how the Doppler Effect works and then be asked to teach the material that you learned. Specifically, you will be expected to provide a short (up to 3 minutes) video-recorded lecture explaining how the Doppler Effect works as if you were teaching the material to someone who has no prior knowledge of the subject. You will have 7 minutes to study the lesson and prepare to teach what you learned. You may take notes on the lesson itself and/or use the blank sheet of paper provided to you by the experimenter; however, you will not be able to use your notes while teaching your lesson. If you have any questions, please ask the experimenter now. He/she will not be able to answer questions once you begin. Two versions of a paper-based lesson on the Doppler Effect (modified from Fiorella & Mayer, 2013) were used: an enhanced version and a base version. Both versions contained identical text, consisting of approximately 350 words separated into four sections: (1) background information about the Doppler Effect, including a real-world example involving an fire truck approaching and passing an observer, (2) a description of the basic characteristics of sound waves (i.e., frequency and wavelength) and how they are related, (3), a explanation of how motion influences sound waves using an analogy of a bug on the surface of a pond, and (4) an explanation of how the analogy relates to the Doppler Effect of sound waves, including the behavior of waves as a sound source approaches and passes by a stationary observer. In the base version of the lesson, the text was presented as four paragraphs corresponding to each of the four parts of the lesson. The base version of the lesson is presented in Appendix A. In the enhanced version of the lesson, instructional features were added to the text that intended to reduce the cognitive demands required to understand the lesson. In particular, the enhanced lesson followed the signaling, segmenting, and multimedia principles of learning from multimedia instruction (e.g., Mayer, 2009). Regarding signaling, the enhanced lesson added a title, subheadings, and numbering to the base version. Regarding segmenting, the enhanced lesson broke down the text of the base

version into more manageable parts. Regarding multimedia, the enhanced version added six drawings to the base version in order to complement information described in the text. Two drawings showed the difference between low and high frequency sound waves and how they differ in wavelength (Part 2), two drawings showed the differences in behavior of water waves for a stationary and moving bug on the surface of a pond (Part 3), and two drawings showed the differences in behavior of sound waves as a sound source approaches and passes by an observer (Part 4). The enhanced version of the lesson is presented in Appendix B. The comprehension test consisted six free-response questions designed to assess participants’ ability to explain key concepts related to the Doppler Effect (i.e., each question targeted the comprehension level of Bloom’s Taxonomy; see Anderson et al., 2001): (1) ‘‘Explain how the Doppler Effect works’’ (2) ‘‘Imagine a fire truck with its siren blaring is approaching an observer standing on a street corner. In this scenario, what would intensify the Doppler Effect? Explain your answer,’’ (3) ‘‘Imagine a fire truck is driving down the road with its siren blaring. An observer in a car nearby who can hear the siren does not experience the Doppler Effect. Why not?’’ (4) ‘‘Imagine a fire truck is driving down the road with its siren blaring. How does the truck’s motion influence the way the driver experiences the sound of the siren? Explain your answer,’’ (5) ‘‘Would the Doppler Effect occur if an observer was approaching a stationary sound source? Explain your answer,’’ and (6) ‘‘Imagine a sound source is approaching a stationary observer. How does the observer perceive the sound at the exact moment when the source crosses paths with the observer? Explain your answer.’’ For Question 1, participants were given 3 min to write an answer; pilot testing indicated that sufficient time was provided to write a response. This item was scored by awarding one point for each of thirteen possible components of the explanation abstracted from the lesson. For example, participants earned points for describing how motion changes the way sound is perceived by an observer, for correctly defining wavelength and frequency, for describing the relationship between frequency, wavelength, and pitch, and for describing how the motion of a sound source (as it approaches and passes by) influences the behavior of sound waves and the way in which the sound is perceived by an observer. For Questions 2–6, participants were given 2 min to answer each question; again pilot testing indicated that sufficient time was provided to write a response to each question. One point was assigned for correctly responding to the question, and two points were assigned for providing a correct explanation of the answer. For example, regarding Question 2, a correct response may include increasing the velocity of the sound source toward the observer, and a correct explanation may include describing how it would increase the frequency of the sound waves between the source and the observer, resulting in a higher perceived pitch. In total, the comprehension test was scored out of a possible 28 points. The post-questionnaire was used primarily for exploratory purposes and asked participants to report how much they agreed with each of seven statements on a seven-point scale ranging from 1 (strongly disagree) to 7 (strongly agree): ‘‘I felt the subject matter was difficult,’’ ‘‘I enjoyed learning about the Doppler Effect,’’ ‘‘I would like to learn this way in the future,’’ ‘‘I feel like I have a good understanding of how the Doppler Effect works,’’ ‘‘After this lesson, I would be interested in learning more about the Doppler Effect,’’ ‘‘I found the lesson about the Doppler Effect to be useful to me,’’ and ‘‘I felt stressed while I was learning about the Doppler Effect.’’ It also asked participants to rate the amount of mental effort they invested while learning about the Doppler Effect on a scale ranging from 1 (very low effort) to 7 (very high effort). Finally, the postquestionnaire asked participants to write any additional comments they had about the study.

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85

2.1.3. Procedure Participants were randomly assigned to a treatment group. There were up to two participants in each session. First, the experimenter provided a brief verbal introduction to the experiment, passed out the consent form for participants to sign, and collected the signed consent forms. Second, the experimenter passed out the demographics questionnaire and collected them when the participants were finished. Third, the experimenter passed out the appropriate instructions sheet. Participants in the test conditions were informed that they would have 7 min to study a lesson on the Doppler Effect and then would be expected to answer questions on the material. Participants in the teach conditions were informed that they would have 7 min to study a lesson on the Doppler Effect and then would be expected to teach the material. Specifically, they were told they would be expected to provide a brief lecture of up to 3 min explaining how the Doppler Effect works as if they were teaching the material to someone who has no prior knowledge of the subject. For sessions that involved two participants, they were told that each would present their lesson individually while the other participant waited in the hallway outside of the laboratory. All participants were informed they could take notes on the lesson itself and on a blank sheet of paper, which was provided to them; however, they were also informed that they could not use their notes while answering questions or teaching. Following instructions, participants were either given the enhanced or base version of the Doppler Effect lesson to study for 7 min. After the study period, all participants completed the comprehension test. Participants who were expecting to teach were told that before teaching their lesson, they would be answering some questions on the material they studied. Each question on the comprehension test was timed and completed individually; 3 min were provided for the first question, and 2 min were provided for each of the remaining five questions. After completing the tests, participants then completed the post-questionnaire. Finally, all participants were debriefed and thanked for their participation. The total duration of the experiment was approximately 30 min. 2.1.4. Results The main purpose of Experiment 1 was to directly test whether preparing to teach helps students better organize the contents of the lesson during learning when compared to preparing for a test. Thus, a 2  2 analysis of variance (ANOVA) was conducted, with expectancy (i.e., prepare to teach or prepare for a test) and lesson format (i.e., base or enhanced) serving as between-subjects factors, and immediate comprehension test scores serving as the dependent measure. Table 1 presents the means and standard deviations of each group on the immediate comprehension test. Two raters scored each of the responses on the comprehension test, yielding high reliability (a = .91). Consistent with predictions, the analysis revealed a significant main effect of expectation, F(1, 94) = 6.72, p = .01, with those expecting to teach outperforming those expecting to be tested (d = .55). However, results revealed no significant main effect or interaction regarding lesson format. Thus, students who prepared to teach outperformed students who prepared for a test regardless of whether they studied the base or enhanced version of the lesson. Table 1 Mean comprehension score (and SD) for Experiment 1 (immediate test). Expectation

Comprehension score Lesson format Base

Test Teach

Enhanced

Overall

n

M

SD

n

M

SD

n

M

SD

23 24

7.8 9.1

3.4 4.9

24 24

7.9 11.0

3.5 4.8

47 48

7.8 10.1*

3.4 4.8

Asterisk (*) indicates significant difference at p < .05, d = .55.

79

As supplementary data, we also considered the effects of teaching expectancy on participants’ levels of motivation, perceived difficulty, and perceived stress during learning. A 2  2 MANOVA was conducted, with expectation and lesson format again serving as between-subjects factors, and each of the eight items of the postquestionnaire serving as the dependent measures. Results indicated several significant main effects of expectation regarding motivation to learn, such that students who prepared to teach reported investing more effort during learning, F(1, 94) = 11.69, p = .001, d = .77, a greater level of enjoyment during learning, F(1, 94) = 8.32, p = .005, d = .56, a greater preference to learn the same way in the future, F(1, 94) = 5.72, p = .019, d = .53, a greater desire to learn more about the material, F(1, 94) = 11.69, p = .001, d = .75, and greater perceived usefulness of the material, F(1, 94) = 6.39, p = .013, d = .58, compared to students who prepared for a test. Students who prepared to teach also reported experiencing significantly higher levels of stress during learning than students who prepared for a test F(1, 94) = 26.68, p < .001, d = 1.03. No main effect for expectation was found regarding the perceived difficulty of the lesson, F(1, 94) = .04, p = .852, d = .00, or their perceived understanding of the material, F(1, 94) = 2.57, p = .112, d = .30. Finally, there were no significant main effects or interactions regarding lesson format for any of the items on the post-questionnaire. 3. Experiment 2 The primary goal of Experiment 2 was to determine if the teaching effect is strongest when students prepare to teach before teaching compared to when students prepare for a test before teaching. 3.1. Method 3.1.1. Participants and design The participants were 104 undergraduate students recruited from the Psychology Subject Pool at a western university. The mean age was 18.5 years (SD = 1.25), and there were 28 men and 76 women. The groups did not differ significantly in terms of mean age or proportion of men and women. The study followed a 2  2 design, in which expectation (i.e., expect teach or expect test) and teaching (i.e., teach or no teach) served as between-subjects factors. Fifty-three participants expected to teach the material following studying (expect teach group), whereas 51 expected to be tested on the material (expect test group). Of those expecting to teach, 27 actually taught the material following studying (expect teach—teach group), whereas 24 only prepared to teach (expect teach—no teach group). Of those expecting to be tested, 27 studied the material without later teaching (expect test—no teach), whereas 26 studied the lesson and then taught the material (expect test—teach). Participants had relatively low prior knowledge of the Doppler Effect, as reported on a questionnaire with a possible score of 13 (expect teach—teach: M = 4.0, SD = 1.8; expect teach—no teach: M = 5.0, SD = 2.4; expect test—teach: M = 4.2, SD = 2.6; expect test—no teach: M = 3.9, SD = 2.0). The groups did not differ significantly on level of prior knowledge. 3.1.2. Materials The consent form, demographics form, comprehension test, and post-questionnaire were they same as in Experiment 1. The only paper-based materials that differed from Experiment 1 were the instructions sheets and the Doppler Effect lesson. There were two instructions sheets given to participants depending on which expectation condition they were randomly assigned. The expect test groups were told that they would be expected to answer questions on the material they studied:

80

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85

In this experiment, you will study a short lesson on how the Doppler Effect works and then be asked to answer some questions about what you learned. You will have 15 minutes to study the lesson. You may take notes on the lesson itself and/or use the blank sheet of paper provided to you by the experimenter; however, you will not be able to use your notes while answering questions on the material. Please continue working until your time has expired. If you have any questions, please ask the experimenter now. He/she will not be able to answer questions once you begin. Participants assigned to one of the expect teach groups were told that they would be expected to teach the material they studied: In this experiment, you will study a short lesson on how the Doppler Effect works and then be asked to teach the material that you learned. Specifically, you will be expected to provide a short (up to 5 minutes) video-recorded lecture explaining how the Doppler Effect works as if you were teaching the material to someone who has no prior knowledge of the subject. You will have 10 minutes to study the lesson and prepare to teach what you learned. You may take notes on the lesson itself and/or use the blank sheet of paper provided to you by the experimenter; however, you will not be able to use your notes while teaching your lesson. If you have any questions, please ask the experimenter now. He/she will not be able to answer questions once you begin. The Doppler Effect lesson was an expanded version of the enhanced lesson used in Experiment 1 and that has also been used in previous work (Fiorella & Mayer, 2013). The lesson consisted of two single-sided pages of both illustrations and text. The text was single-spaced and consisted of approximately 600 words (whereas the lesson used in Experiment 1 consisted of approximately 350 words). Similar to Experiment 1, the lesson was divided into four main parts. First, a concrete example of the Doppler Effect was provided in which the reader was asked to imagine a fire-truck approaching and passing by them. The siren’s perceived pitch was described as getting higher as the fire truck gets closer to them and lower as the fire truck moves further away. Second, a description of the basic characteristics of sound waves was provided, including a definition and illustration of wave frequency and wavelength. Third, an analogy was described comparing the Doppler Effect to a bug jiggling on the surface of a pond. Specifically, if the bug is stationary, the water waves it produces will occur at the same frequency and wavelength in all directions; if the bug moves to the right, the waves will occur more frequently to the right of the bug (i.e., in the direction of the bug’s movement) and less frequently to the left of the bug. Finally, the lesson returned to the original scenario of an approaching fire truck and explained how the fire truck’s motion influences the behavior of sound waves and how as a result, sound is perceived differently as the fire truck approaches and as it passes an observer. The lesson used in Experiment 2 is presented in Appendix C. The apparatus consisted of one Cannon FS400 camcorder used to record the lectures given by participants assigned to teach groups. 3.1.3. Procedure Participants were randomly assigned to a treatment group. There were one or two participants in each session. First, the experimenter provided a brief verbal introduction to the experiment, passed out the consent form for participants to sign, and collected the signed consent forms. Second, the experimenter passed out the demographics questionnaire and collected them when the participants were finished. Third, the experimenter passed out the appropriate instructions sheet.

Participants assigned to the expect test—no teach group were informed that they would have 15 min to study a lesson on the Doppler Effect and then would be expected to answer questions on the material, whereas participants assigned to the expect test—teach group were informed that they would have 10 min to study the lesson and then would answer questions on the material. Participants assigned to the expect teach—no teach group were informed that they would have 15 min to study the lesson and then would have up to 5 min to provide a video recorded lecture as if they were teaching the contents of the lesson to another person who has no prior knowledge of the subject. Finally, participants in the expect teach—teach group were informed that they would have 10 min to study the lesson before teaching the material for up to 5 min. Thus, this procedure ensured that all groups received 15 min either studying or teaching the material. All groups were informed they could take notes on the lesson itself and that they could take notes on a blank sheet of paper, which was provided to them. Following the study period, the experimenter collected the lesson and any notes taken by participants. Participants assigned to no teach conditions (who studied the lesson for 15 min) were then told that they had completed Part 1 of the experiment and were reminded to return for Part 2 of the experiment in 1 week. Participants were not provided with any details regarding the tasks involved in Part 2. Participants assigned to teach conditions (who studied the lesson for 10 min) were given up to 5 min to provide a video-recorded lecture of the lesson’s contents as if they were teaching someone who has no prior knowledge of the subject. In cases when two participants were in a session, a coin toss decided which participant would give their lecture first. The other participant was asked to wait in the hallway outside of the laboratory. The participants then switched positions for the second participant to provide their lecture. After participants completed their teaching task, they were told that they had completed Part 1 of the experiment and were reminded to return for Part 2 of the experiment in 1 week. Participants were not provided with any details regarding the tasks involved in Part 2. One week after completing Part 1 of the experiment, participants returned to complete Part 2. First, participants were given the comprehension test. Each question on the comprehension test was timed and completed individually; 3 min were provided for the first question, and 2 min were provided for each of the remaining five questions. After completing the comprehension test, participants then completed the post-questionnaire. Finally, all participants were debriefed and thanked for their participation. The total duration of Part 1 was approximately 25 min, and the total duration of Part 2 was approximately 20 min. 3.1.4. Results The main goal of Experiment 2 was to test the extent to which long-term effects of teaching depended on the expectation students were given before studying (i.e., to teach or to take a test). Table 2 reports the means and standard deviations of each group on the delayed comprehension test. Two raters again scored each response on the comprehension test, yielding high reliability (a = .94). Overall, it was predicted that there would be a main effect of teaching, such that those who actually teach the material outperform those who do not teach the material. Further, since performance was measured on a delayed test, it was predicted that there would be no main effect of expectation. Finally, it was predicted that there would be an interaction between expectation (i.e., test or teach) and activity (i.e., teach or no teach) such that the benefits of actually teaching are greatest for those who also prepared to teach. To test these hypotheses, a 2  2 factorial ANOVA was conducted, with expectation and activity serving as between-subjects factors and performance on the delayed comprehension test

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85 Table 2 Mean Comprehension Score (and SD) for Four Groups in Experiment 2 (Delayed Test). Expectation

Activity

Test Teach Test Teach

No Teach No Teach Teach Teach

Comprehension score n

M

SD

24 27 27 26

5.0 4.1 5.8 7.0*

3.7 2.9 3.5 3.5

Effect size (d)

.27 .22 .56

Asterisk (*) indicates significantly different from Expect Test—No Teach Group at p < .05.

serving as the dependent measure. Consistent with predictions, the analysis revealed a main effect of teaching, F(1, 103) = 7.83, p < .01, with those who taught the material outperforming those who did not teach the material (d = .56), and no main effect of expectation, F(1, 103) = .06, p = .80. Further, the analysis revealed a significant interaction between expectation and activity, F(1, 103) = 4.35, p < .05, such that those who prepared to teach and actually taught performed better on the comprehension test than the control group (d = .56), whereas those who prepared to teach and did not actually teach performed worse on the comprehension test than the control group (d = .27). Although students who taught the material, in general, outperformed those who did not teach the material, the interaction reflects a pattern in which students who taught without the expectation of teaching showed a small teaching effect as compared to the control group (d = .22), whereas students who taught with the expectation of teaching showed a medium teaching effect (d = .55). A one-way ANOVA across the four groups was also significant, F(3, 100) = 3.45, p < .05, and Dunnett post hoc tests at p < .05 indicated that only those who prepared to teach and actually taught the material significantly outperformed the control group. Overall, these results suggest the act of teaching coupled with the expectation of teaching is important for ensuring long-term comprehension, whereas the expectation of teaching alone is insufficient for ensuring long-term comprehension. As in Experiment 1, we also considered the supplementary data on students’ self-reported motivation to learn, perceived difficulty, and perceived stress during learning. A 2  2 MANOVA was conducted, with expectation and activity again serving as betweensubjects factors and each of the eight items on the post-questionnaire serving as dependent measures. In contrast to Experiment 1, no significant main effects or interactions were found across conditions for the items of the post-questionnaire. However, students completed the post-questionnaire at the end of the experiment after the delayed comprehension test. Thus, students may not have been able to retrospectively report their level of motivation, perceived difficulty, and stress during initial learning, which took place 1 week previously. 4. Discussion 4.1. Empirical contribution The current study provides further empirical support for a teaching expectancy effect for immediate learning (Experiment 1; d = .55) and a teaching effect for long-term learning (Experiment 2; d = .56). In Experiment 1, the teaching expectancy effect was not influenced by whether students learned from the base version or the enhanced version of the lesson. Further, supplementary self-report data from the post-questionnaire suggests that students who prepared to teach (regardless of lesson format) may have been more motivated to invest effort during learning than those who prepared for a test of the material. Taken together, these findings suggest that preparing to teach results in greater short-term

81

learning benefits compared to preparing for a test of the material, which may be in part because students are more motivated to learn the material. In Experiment 2, the teaching effect was strongest when students both prepared to teach and actually taught the material (d = .56) compared to those who prepared for a test and then taught the material (d = .22). Further, preparing to teach (without actually teaching) did not enhance learning on the delayed comprehension test, consistent with past research on the teaching expectancy effect (Fiorella & Mayer, 2013). Taken together, these findings suggest that preparing to teach—while on its own is insufficient to promote long-term learning effects—may still play an important role in promoting long-term learning when it is coupled with actually teaching. 4.2. Theoretical contribution On the theoretical level, these findings suggest that the act of teaching (i.e., explaining material to others) may promote generative processing necessary for long-term meaningful learning. According to the cognitive theory of multimedia learning (Mayer, 2005, 2009, 2011), generative processing involves actively constructing a coherent representation of the material that fits with learners’ prior knowledge, by engaging in the cognitive processes of organizing and integrating. Although the current study does not include direct measures of generative processing, performance on the comprehension test suggests that students who actually taught the contents of the lesson were better able to make sense out of the material than students who did only prepared to teach or prepared for a test. Thus, the act of teaching may encourage learners to invest effort toward learning for understanding, presumably by better organizing and integrating the material with their existing knowledge. The benefits of actually teaching appear to depend somewhat on whether students also prepare to teach. The current study suggests that the mere expectation of teaching may not prime learners to engage in deep cognitive processing that supports long-term comprehension; rather, students may simply invest more effort using their own existing strategies aimed at short-term performance. This provides one potential explanation for why preparing to teach only leads to short-term learning benefits (as indicated in Experiment 1) but not long-term learning benefits (as indicated in Experiment 2). At the same time, the current study suggests that preparing to teach plays an important role in long-term learning when it is coupled with actually teaching. 4.3. Practical contribution On the practical level, the current study suggests that explaining recently learned material to others provides students with a learning strategy that promotes meaningful learning over time. This is consistent with the idea that long-term learning is enhanced when students employ learning strategies that encourage them to actively make sense out of the material by making connections between the to-be-learned material and their prior knowledge—similar to other learning strategies such as self-testing (e.g., Roediger & Karpicke, 2006) or self-explanation (e.g., Chi, de Leeuw, Chiu, & LaVancher, 1994). The current study is also consistent with the idea that practice in using the material by accessing recently learned material is critical for long-term learning and more effective than extended studying—similar to ideas of retrieval-based learning (Karpicke, 2012). Finally, the current study suggests that teaching can be used as an effective learning strategy even without training and without actually interacting with another student.

82

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85

4.4. Limitations and future directions One limitation of the current study is that the wording of the text of the base version of the lesson in Experiment 1 may have been organized and concise enough to not require additional instructional features (e.g., signaling, segmenting, and multimedia) used in enhanced version. Similarly, since participants were able to take notes during learning, any detrimental effects of studying the base version of the lesson may have been mediated. Future research is needed to determine whether teaching expectancy prompts are particularly effective for extremely poorly designed lessons or when students are unable to take notes during learning. Further research should also explore whether there are ways in which preparing to teach can better contribute to the long-term benefits of learning by teaching. For example, it may be that students need more guidance in how to prepare to teach or that they need to be prompted to engage in more specific activities related to preparing to teach (e.g., creating a lesson plan). In other words, simply giving students the expectation of later teaching may not sufficiently prime the use of effective learning strategies that promote long-term learning. The present study did not include direct measures of cognitive processing during learning; thus, generative processing was inferred based on performance on the comprehension test. Additional research should include more process-based data to provide deeper insight into the cognitive mechanisms underlying preparing to teach and explaining to others. These more fine-grained direct measures of cognitive processing during learning include note-taking behavior or eye tracking while preparing to teach (versus preparing for a test), or analysis of the quality of explanations provided while actually teaching. Similarly, future research on learning by teaching may benefit from the inclusion of more direct measures of cognitive load and motivation during learning. Finally, research is needed to pinpoint the specific features of teaching that promote learning. For example, in Experiment 2 students who taught the material did so by standing and explaining the material aloud in front of a video camera. They were also free to use gestures as they provided their explanation. Although this was done to provide students with a relatively authentic teaching task, it is unclear how each of these features differentially contributed to comprehension performance. On the other hand, teaching may be effective because it simultaneously incorporates many features that potentially result in enhanced learning (e.g., speaking aloud, explaining, gesturing). Future research should explore the extent to which effects of actually teaching may depend on these different features. Acknowledgment This research was supported by Grant N000140810018 from the Office of Naval Research.

waves. Frequency is the number of waves passing through a given point during a period of time. Longer sound waves require more time to travel a given distance, so they occur less frequently and have a lower pitch. Shorter sound waves require less time to travel a given distance, so they occur more frequently and have a higher pitch. Motion influences the frequency and wavelength of waves. To illustrate, imagine a bug jiggling on the surface of a pond. If the bug is stationary, the waves on the surface of the water around it will be at the same frequency and length in all directions. When the bug moves to the right, the waves it produces become shorter and more frequent to the right of the bug and longer and less frequent to the left of the bug. To relate the bug analogy to the Doppler Effect, imagine that a fire truck is approaching an observer with its siren blaring. As the fire truck approaches, the observer receives the sound waves at a higher frequency. Specifically, the sound waves take less time to reach the observer because they are being compressed closer together between the fire truck and the observer. As a result, the observer experiences an increasing pitch. As the fire truck passes by, the observer receives the waves at a lower frequency. Specifically, the sound waves take longer to reach the observer because they are being stretched further apart between the fire truck and the observer. As a result, the observer experiences a decreasing pitch. Appendix B. Enhanced version of lesson in Experiment 1 B.1. The Doppler Effect B.1.1. Background The Doppler Effect is very common, though many people do not know what causes it. For example, imagine a person is standing on a street corner as a fire truck approaches with its siren blaring: The perceived pitch of the siren will get higher as it comes closer to the observer; as it passes by, the observer will perceive the pitch as getting lower. In order to understand how this process works, it is important to first understand some of the basic characteristics of sound waves. B.1.2. Characteristics of sound waves Sound waves have two primary characteristics: wavelength and frequency. Wavelength is the distance between adjacent waves. Frequency is the number of waves passing through a given point during a period of time. Longer sound waves require more time to travel a given distance, so they occur less frequently and have a lower pitch.

Wavelength

Appendix A. Base version of lesson in Experiment 1 The Doppler Effect is very common, though many people do not know what causes it. For example, imagine a person is standing on a street corner as a fire truck approaches with its siren blaring: The perceived pitch of the siren will sound higher to the observer as it approaches; as it passes by, the observer will perceive the pitch as getting lower. In order to understand how this process works, it is important to first understand some of the basic characteristics of sound waves. Sound waves have two primary characteristics: wavelength and frequency. Wavelength is the distance between adjacent

Low Frequency Waves Shorter sound waves require less time to travel a given distance, so they occur more frequently and have a higher pitch.

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85

Wavelength

83

As the fire truck passes by, the observer receives the waves at a lower frequency. Specifically, the sound waves take longer to reach the observer because they are being stretched further apart between the fire truck and the observer. As a result, the observer experiences a decreasing pitch

High Frequency Waves B.1.3. Motion and waves Motion influences the frequency and wavelength of waves. To illustrate, imagine a bug jiggling on the surface of a pond: If the bug is stationary, the waves on the surface of the water around it will be at the same frequency and length in all directions.

Fire Truck Passes By Appendix C. Expanded version of lesson in Experiment 2 C.1. The Doppler Effect

Stationary Bug When the bug moves to the right, waves it produces become shorter and more frequent to the right of the bug and longer and less frequent to the left of the bug.

C.1.1. Background Almost everyone has experienced the Doppler Effect, though perhaps without knowing what caused it. For example, imagine you are standing on a street corner as a fire truck approaches with its siren blaring. The perceived pitch of the siren will sound higher as it comes closer to you. Then, as it passes by, the pitch will sound lower. This is one of many examples of the Doppler Effect: the change in how sounds are perceived due to movement. C.1.2. Sound waves Why does this change occur? Movement changes the way different characteristics of sound waves are perceived, and therefore, how the sound is perceived. Sound waves have two primary characteristics: frequency and wavelength. As we will see, movement causes changes in how we perceive the frequency and length of sound waves, which ultimately impact how we perceive the sound. First, let’s briefly go over each of these characteristics.

Bug Moving to the Right B.1.4. How the Doppler Effect works To relate the bug analogy to the Doppler Effect, imagine that a fire truck is approaching an observer with its siren blaring. As the fire truck approaches, the observer receives the sound waves at a higher frequency. Specifically, the sound waves take less time to reach the observer because they are being compressed closer together between the fire truck and the observer. As a result, the observer experiences an increasing pitch

Fire Truck Approaching

C.1.3. Wave frequency Wave frequency refers to the number of waves passing through a given point during one second. It corresponds to how we perceive the pitch of a sound: if waves occur at a high frequency, they will produce a high pitch; if waves occur at a low frequency, they will produce a low pitch. For example, the cry of a baby has a relatively high pitch, while the sound of thunder has a relatively low pitch. The reason these two sounds are perceived differently is because they have different wave frequencies. Figs. 1 and 2 illustrate the difference between low and high frequency sound waves.

Fig. 1. Low frequency sound waves.

84

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85

Fig. 2. High frequency sound waves.

C.1.4. Wavelength Closely related to wave frequency is wavelength. Wavelength refers to the distance between adjacent waves (see figures above). As you might expect, longer waves require more time to travel a given distance than shorter waves. Consequently, longer sound waves have a lower frequency and a lower pitch. On the other hand, short sound waves have a higher frequency and higher pitch.

C.1.5. How the Doppler Effect works The Doppler Effect is about how movement influences how the frequency and length of sound waves are perceived. To illustrate this, imagine a bug jiggling on the surface of a pond. If the bug is stationary, the waves on the surface of the water around it will be at the same frequency and length in all directions, as in Fig. 3. Now suppose that the bug begins moving to the right. The waves it produces become shorter and more frequent to the right of the bug and longer and less frequent to the left of the bug, as shown in Fig. 4.

Fig. 4. Bug moving to the right.

Now let’s relate the bug example to how the Doppler Effect occurs in sound waves. Imagine again that a fire truck is approaching with its siren blaring, as illustrated in Fig. 5. As the fire truck approaches, the sound waves between the fire truck and the girl become shorter and more frequent, resulting in a higher perceived pitch. As the ambulance drives by, the sound waves between the fire truck and the girl are longer and less frequent. As a result, the girl perceives the pitch as getting lower. This is because the movement of the fire truck causes changes in how the sound is perceived. This influence of movement on perceived sound is the core principle of the Doppler Effect.

References

Fig. 3. Stationary bug.

Allen, V., & Feldman, R. (1973). Learning through tutoring: Low-achieving children as tutors. Journal of Experimental Education, 42(1), 1–5. Anderson, L. W., Krathwohl, D. R., Airasian, P. W., Cruikshank, K. A., Mayer, R. E., Pintrich, P. R., et al. (2001). A taxonomy for learning, teaching, and assessing: A revision of Bloom’s taxonomy of educational objectives. New York: Longman. Annis, L. F. (1983). The processes and effects of peer tutoring. Human Learning: Journal of Practical Research & Applications, 2(1), 39–47. Bargh, J. A., & Schul, Y. (1980). On the cognitive benefits of teaching. Journal of Educational Psychology, 72, 593–604. Benware, C. A., & Deci, E. L. (1984). Quality of learning with an active versus passive motivational set. American Educational Research Journal, 21, 755–765. Biswas, G., Leelawong, K., Schwartz, D., Vye & The Teachable Agents Group at Vanderbilt (2005). Learning by teaching: A new agent paradigm for educational software. Applied Artificial Intelligence, 19, 363–392. Chi, M. T. H. (2000). Self-explaining expository texts: The dual processes of generating inferences and repairing mental models. In R. Glaser (Ed.), Advances in Instructional Psychology (pp. 161–238). Mahwah, NJ: Lawrence Erlbaum Associates. Chi, M. T. H., de Leeuw, N., Chiu, M., & LaVancher, C. (1994). Eliciting selfexplanations improves understanding. Cognitive Science, 18, 439–477. Cohen, P., Kulik, J., & Kulik, C. (1982). Educational outcomes of tutoring: A metaanalysis of findings. American Educational Research Journal, 19(2), 237–248.

Fig. 5. The Doppler Effect of sound waves.

L. Fiorella, R.E. Mayer / Contemporary Educational Psychology 39 (2014) 75–85 Coleman, E. B., Brown, A. L., & Rivkin, I. D. (1997). The effect of instructional explanations on learning from scientific texts. The Journal of the Learning Sciences, 6(4), 347–365. Ehly, S., Keith, T. Z., & Bratton, B. (1987). The benefits of tutoring: An exploration of expectancy and outcomes. Contemporary Educational Psychology, 12, 131–134. Fiorella, L., & Mayer, R. E. (2013). The relative benefits of learning by teaching and teaching expectancy. Contemporary Educational Psychology, 38, 281–288. Fonseca, B. A., & Chi, M. T. H. (2011). Instruction based on self-explanation. In R. E. Mayer & P. A. Alexander (Eds.), Handbook of research in learning and instruction (pp. 296–319). New York, NY: Routledge. Galbraith, J., & Winterbottom, M. (2011). Peer tutoring: What’s in it for the tutor? Educational Studies, 37(3), 321–332. Gartner, A., Kohler, M. C., & Riessman, F. (1971). Children teach children: Learning by teaching. New York: Harper & Row. Karpicke, J. D. (2012). Retrieval-based learning: Active retrieval promotes meaningful learning. Current Directions in Psychological Science, 21(3), 157–163. King, A., Staffieri, A., & Adelgais, A. (1998). Mutual peer tutoring: Effects of structuring tutorial interaction to scaffold peer learning. Journal of Educational Psychology, 90(1), 134–152. Mayer, R. E. (2005). Cognitive theory of multimedia learning. In R. E. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 31–48). New York, NY: Cambridge University Press. Mayer, R. E. (2009). Multimedia learning (2nd ed.). New York, NY: Cambridge University Press. Mayer, R. E. (2011). Applying the science of learning. Upper Saddle River, NJ: Pearson Merrill Prentice Hall. Morgan, R., & Toy, T. (1970). Learning by teaching: A student-to student compensatory tutoring program in a rural school system and its relevance to the educational cooperative. Psychological Record, 20, 159–169.

85

Palincsar, A. S., & Brown, A. L. (1984). Reciprocal teaching of comprehensionfostering and comprehension-monitoring activities. Cognition and Instruction, 1, 117–175. Renkl, A. (1995). Learning for later teaching: An exploration of meditational links between teaching expectancy and learning results. Learning and Instruction, 5, 21–36. Robinson, D., Schoefield, J., & Steers-Wentzell, K. (2005). Peer and cross-age tutoring in math: Outcomes and their design implications. Educational Psychology Review, 17(4), 327–362. Roediger, H. L., (III) & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17, 249–255. Rohrbeck, C., Ginsburg-Block, M., Fantuzzo, J., & Miller, T. (2003). Peer-assisted learning interventions with elementary school students: A meta-analytic review. Journal of Educational Psychology, 95(2), 240–257. Roscoe, R. D., & Chi, M. T. H. (2007). Understanding tutor learning: Knowledgebuilding and knowledge-telling in peer tutors’ explanations and questions. Review of Educational Research, 77(4), 534–574. Roscoe, R. D., & Chi, M. T. H. (2008). Tutoring learning: The role of explaining and responding to questions. Instructional Science, 36, 321–350. Roy, M., & Chi, M. T. H. (2005). The self-explanation principle in multimedia learning. In R. E. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 271–286). New York: Cambridge University Press. Slavin, R. E. (1983). Cooperative learning. New York: Longman. Sweller, J. (2005). Implications of cognitive load theory for multimedia learning. In R. E. Mayer (Ed.), The Cambridge handbook of multimedia learning (pp. 19–30). New York: Cambridge University Press. Sweller, J., Ayres, P., & Kalyuga, S. (2011). Cognitive load theory. New York: Springer.