1 Introduction

Instructional pacing is one of the most substantial challenges teachers face in effectively meeting the needs of all their students. Typically defined as the speed or rate at which a teacher presents a task and/or information, pacing has been shown to be an important factor in student learning. Appropriately paced lessons have many benefits, including enhancing student attention and increasing the number of response opportunities. Inappropriately paced lessons, on the other hand, can lead to classroom management issues and poor learning outcomes. A number of strategies for effective pacing have been identified, including providing clear goals, articulating expectations, embedding smooth transitions between activities and topics, and preparing all lesson materials are ready beforehand. However, many teachers report this aspect of teaching is a continued challenge and a particularly difficult one with direct instruction. Developing an appropriately paced lesson that meets the needs of all students is a daunting task, yet one that is prevalent in traditional classrooms that rely on direct instruction. How can direct instruction, such as lectures, meet the individual needs of all students in a classroom?

In response to this concern, a pedagogical approach has emerged that flips the traditional classroom experience. Bergmann and Sams (2012), high school Chemistry teachers who coined “flipping a classroom”, described this pedagogical approach as what is, “…traditionally done in class is now done at home, and that which is traditionally done as home is now completed in class” (Bergmann and Sams 2012, p. 13). The challenges instructors face when lecturing during class can be minimized in an effectively designed flipped classroom. Lectures in more traditional based courses are generally aimed at the middle performing students, with effect of leaving the higher performing students bored and the lower performing students frustrated. Flipped classrooms, on the other and, address this issue by offering a more individualized educational experience through recorded lectures assigned as homework. Unlike learning from traditional lectures offered during class, videos designed for flipped classrooms provide students with the opportunity to control the pacing and sequencing of information. In a way, this approach offers a mastery approach to learning. Students who are struggling with the content can pause and replay the video as many times as necessary. Students who have mastered material, on the other hand, can spend less time viewing it. Thus, content delivery is more highly individualized in a flipped classroom. Furthermore, assigning lecture videos as homework allows a teacher to redistribute precious class time. Additional time becomes available for collaborative discussions and application activities that may not otherwise be possible in traditional classrooms. While the term “flipped classroom” is relatively new, the techniques guiding this pedagogical approach are drawn from well research theories, such as Constructivism, and established instructional practices (Bergmann and Sams 2012). The uniqueness of the flipped classroom comes in the form of its delivery, which typically utilizes videotaped lectures assigned as homework. The increasing accessibility of technologies to students and recent advancements that make high-quality video even easier to create offer a fairly unique application of established approaches to teaching and learning. Given the theoretical foundation of the flipped classroom and recent technological advancements, it is not surprising that this model is growing in popularity on college campuses. The flipped classroom model has even been recognized by EDUCAUSE as an emerging technology-based strategy in higher education. Despite the increasing prevalence of this model, however, there is limited empirical data that supports the pedagogical approach of flipping the classroom. Existing empirical research tends to be mixed and peer review is typically anecdotal (Herreid and Schiller 2013). In particular, there is a lack of research that has used process data to examine how students learn with videos designed for flipped classrooms. In the absence of this research, there is little to guide the actual design of videos for this classroom model.

1.1 Learning with Hypermedia

One of the potential benefits of using videos in the flipped classroom model is the nonlinear access to information. While watching a videotaped lesson, students can control the sequencing of information by pausing, rewinding, and watching the video at a pace that best meets their cognitive needs. Furthermore, technology advancement has made it even easier to create videos that integrate multiple representations of information (audio with video and animation, for example). Hypermedia, defined as a non-linear computer based environment that integrates text with audio, video, animation, and/or graphics (Jacobson and Archodidou 2000), shares many of the same design features as videos designed for flipped classroom. When learning with hypermedia, students can access multiple representations of information in a variety of sequences through hyperlinked nodes, such as search engines and hyperlinks in a webpage. This non-linear presentation of information allows students to freely navigate and determine the sequence of information. While the educational community initially embraced hypermedia, research has revealed inherent challenges created by the very features designed to facilitate learning. Learning with hypermedia places cognitive and metacognitive demands on the student. For example, students need to determine how much time to spend in different representations of information (Azevedo 2014; Azevedo et al. 2010; Shapiro 2008). Furthermore, non-linear access requires students to monitor the relevancy of information in inter-linked nodes (Azevedo 2008; Azevedo et al. 2012; Greene and Azevedo 2009; Moos and Stewart 2013; Moos 2014; Tuysuzoglu and Greene 2015; Winne and Nesbit 2009; Zimmerman 2008). This autonomy and access to multiple representations of information require students to monitor comprehension and use repair strategies when comprehension breaks down (Azevedo 2009; Moos and Azevedo 2008; Johnson et al. 2011). Not surprisingly, empirical research has found that certain processes, such as metacognitive monitoring, positively predict learning outcomes with these technology environments (Azevedo 2009; Moos 2011). These processes have been characterized as self-regulated learning (SRL; Pintrich 2000; Schunk and Zimmerman 2013; Winne 2001; Winne and Perry 2000; Winne and Hadwin 1998; Zimmerman 2006, 2008) and research has turned to SRL theories to better understand learning with hypermedia.

1.2 Self-Regulated Learning Theory

Though the field of SRL consists of divergent theoretical perspectives (Boekaerts et al. 2000; Zimmerman 2001), there is agreement on its conceptualization. Broadly speaking, SRL is conceptualized as learning that involves the regulation and monitoring of cognition, behavior, and motivation, and the active construction of knowledge by using strategies and goals (Azevedo et al. 2012; Cleary and Platten 2013; Schunk and Zimmerman 2013; Winne and Hadwin 1998; Winne 2005; Zimmerman 2006, 2008). Underlying this conceptualization are four common assumptions evident in most SRL models (Pintrich 2000). First, it is assumed that students actively construct their own strategies and goals. Second, students can potentially regulate and monitor certain aspects of their cognition, behavior, and motivation. Third, most models assume that all human cognitive behavior is goal-directed and that self-regulated students modify their behavior to achieve a desired goal. Self-regulated students set goals for their learning, monitor progress towards these goals, and then adapt and regulate their behavior, cognition, and motivation to reach those goals. Fourth, most models assume that self-regulatory behavior is a mediator between (a) an individual’s performance, (b) contextual factors, and (c) personal characteristics.

These core assumptions provide the foundation for different theoretical approaches that explain active learning (see Zimmerman and Schunk 2001 for a review), including the widely cited Social-Cognitive Theory of SRL (Zimmerman 2006, 2008; Schunk and Usher 2012; Schunk and Mullen 2012; Schunk and Zimmerman 2013). According to this theory, SRL is comprised of three interactive phases: forethought, performance control, and self-reflection. The first phase of self-regulation entails an analysis of the learning task and subsequent creation of goals, processes that are framed by motivation orientations. Different intellectual traditions have resulted in distinct theoretical perspectives of motivation orientations, which include beliefs, values, and goals (Eccles and Wigfield 2002). Research guided by the Social Cognitive Theory of SRL places particular emphasis on self-efficacy, task value, intrinsic motivation, and control beliefs. Orientations within each of these motivations are thought to influence the next phase of self-regulated learning, the performance phase. The use of strategies during this phase, such as summarizing, taking notes, attention focusing and self-instruction, facilitate performance. The role of metacognitive monitoring also assumes a critical role in this phase. Identifying discrepancies between current knowledge state and the learning task goal provides internally generated feedback that allows students to regulate and govern task execution. The final phase of SRL, self-reflection, occur when students judge and develop reasons for their performance. Self-evaluations in this phase tend to be idiosyncratic because they reflect both the assessment criteria and the student’s individual performance level. Emotions experienced during the self-reflective phase of SRL affect subsequent motivational orientations, variables described in the planning phase in this SRL theoretical framework.

1.3 Current Study

Research has used the SRL theory to examine how students learn with non-linear environments, such as hypermedia. The inherent design issues create cognitive and metacognitive challenges, often requiring students to regularly monitor their emerging understanding, evaluate content while navigating through the nonlinear environment, and coordinate multiple sources of information. Research has found that while the design of hypermedia necessitates these SRL processes, many students (including undergraduates) do not adequately self-regulate their learning (e.g., Moos 2013). In order to address the challenges students face during learning with hypermedia, research has examined the effect of instructional support during learning. Embedded prompts, for example, can direct students to perform specific activities during learning (Wirth 2009). Prompts can be presented in various forms, including incomplete sentences (e.g., “My first step should be to….”), simple questions (“What will your first step be in learning about this topic?”), explicit procedural instructions (“Your first step should be to write a list of questions you have on the topic”), or graphics directing learning (Bannert 2009). Prompts that support self-regulated learning (Ifenthaler 2012) are of particular interest for the flipped classroom models due to the central role of videos. Past research has provided some direction on the types of prompts that successfully lead students to engage in self-regulation during learning within hypermedia contexts. Hoffman and Spatariu (2011), for example, demonstrated that metacognitive prompts during the solving of multiplication problems encouraged students in reflective cognition. More recently, Azevedo and his team have developed MetaTutor, which is an advanced hypermedia learning environment designed to detect, trace, and foster students’ self-regulated learning. The design of this environment is based on extensive research by Azevedo and colleagues (see Azevedo et al. 2009) and capitalizes on adaptive scaffolding to enhance learning. These lines of research provide fairly convincing evidence that students benefit from embedded prompts, particular in non-linear environment such as hypermedia (Bannert and Reimann 2012; Feyzi-Behnagh et al. 2014). In the absence of prompts and other scaffolds, students may not adequately self-regulate learning.

These lines of research suggest that embedded prompts in videos designed for flipped classrooms are necessary in order to maximize learning. In the absence of these prompts, students may not fully engage a video that is assigned as homework. In fact, research has revealed mixed findings on the effectiveness of the flipped classroom model (Enfield 2013; Hashmi and Shih 2013; McLaughlin et al. 2013; Smith 2013), suggesting that some students are not adequately engaged. These mixed results raise important questions, including: To what extent do students self-regulate their learning with videos designed for flipped classrooms? To what extent can embedded prompts in these videos facilitate SRL? Addressing these questions is not simply a process of adding prompts to videos. Rather, embedded SRL prompts need to be theoretically aligned and the instructional efficiency of the re-designed video should be considered. Embedding prompts could potentially impose additional, unproductive mental effort upon the student, thus degrading the instructional efficiency of the video.

In sum, prior research has highlighted the challenges some students face when learning with non-linear environments and the potential benefit of embedded SRL prompts in these environments. While the increasing popular flipped classroom model has received considerable attention, there is little empirical research on how students learn within the non-linear context of videos so often assigned for homework in these models. Furthermore, little is known about how to best design these videos to maximize performance while minimizing mental effort, otherwise known as instructional efficiency. This study sought to address these gaps by extending current findings from the field of hypermedia learning. Specifically, this study was guided by the following research questions:

  1. 1.

    What metacognitive processes predict students’ control of video?

  2. 2.

    To what extent do embedded prompts affect SRL and instructional efficiency?

  3. 3.

    To what extent do embedded prompts affect learning outcomes?

2 Methods

2.1 Participants

The sample included 32 pre-service teachers from a college located in the Midwest of the USA. The small sample size was due to the time intensive nature of individually running participants through the study and coding data from a concurrent think-aloud protocol. This sample included 20 females (63 %) and 12 males (37 %), a gender distribution that is reflected by the overall population in this teacher education program. Their average age was 19.93 (SD = 0.92) and their average grade point average was 3.39 (SD = 0.39).

2.2 Measures

The pretest and posttest were identical essays that assessed participants’ knowledge related to theories of motivation, which aligned with the content in the video used for this study. Participants were scored on their responses related to the specific theories addressed in the video. Participants received 0.5 points for the correct identification of the theory and up to another 1.0 point for accurately identifying the assumptions related to that theory (range 0–7.5 on both the pretest and posttest). “Appendix 1” provides an example pretest from a participant in the control condition and “Appendix 2” offers an example from a participant in the experimental condition.

A concurrent think-aloud protocol (Ericsson 2006; Ericsson and Simon 1993) was used to measure the participants’ SRL with the video. A modified scheme, based on several models of SRL (Azevedo and Cromley 2004; Pintrich 2000; Winne 2005; Winne and Hadwin 1998; Zimmerman 2006), was used to analyze the think-alouds. The scheme included the following macro-level codes (with micro-level codes in parentheses): Planning (prior knowledge activation, recycling goals in working memory, setting a sub-goal), Monitoring (Content Evaluation (+), Content Evaluation (−), Monitoring Understanding (+), Monitoring Understanding (−), Monitoring Progress Toward Goal, Time Monitoring), and Learning Strategies (Pause/restart, Rewind, Inferences, Memorization, Re-reading, Review Notes, Summarization, and Take Notes).

Lastly, participants’ mental effort was measured. Following the completion of the video and posttest, participants were asked to rate their mental effort for learning from the video on a 9-point scale ranging from 1 (extremely low) to 9 (extremely high). This approach to measuring mental effort has been previously used to assess efficiency of multimedia and hypermedia environments (Paas et al. 2003, 2005; van Merriënboer and Sweller 2005).

2.2.1 Technology, Learning Environment, and SRL Prompts

The first author, who has over 10 years of experience teaching an Educational and Cognitive Psychology course to pre-service teachers, created the video for this study. It was designed to be viewed as homework for an undergraduate Educational Psychology course. The video, a narrated powerpoint edited through iMovie, introduced several predominant motivation theories. The video for the control condition was 14 min and 5 s and the video for the experimental condition was 16 min and 12 s. The video in the experimental condition was longer due to the embedded SRL scaffolds, which were designed based on Zimmerman’s Social Cognitive approach self-regulation (Zimmerman 2006, 2008; Schunk and Usher 2012; Schunk and Mullen 2012; Schunk and Zimmerman 2013). This commonly cited theory has been extensively used to describe three phases of self-regulation: forethought/planning, performance/monitoring, and self-reflection. The planning prompts, which were provided at the start of the video for the experimental condition, included the following questions:

  1. 1.

    What do you already know about motivating students?

  2. 2.

    What questions do you have about motivating students?

  3. 3.

    What strategies do you think will be effective while learning about motivation in this video?

The monitoring prompts, which provided approximately halfway through the video, included the following questions:

  1. 1.

    What information have you learned so far?

  2. 2.

    What questions (if any) do you have about the information presented so far and/or is there anything presented so far that you do not understand?

  3. 3.

    How effective have your strategies been in learning about motivation in this video?

  4. 4.

    Do you need to adjust how you are learning?

Lastly, the reflection prompts, which were provided at the end of the video, included the following questions:

  1. 1.

    What did you learn about motivating students?

  2. 2.

    What questions (if any) do you have about the information presented and/or is there anything that you did not understand in the video?

  3. 3.

    Do you need to go back in the video and fill any gaps in understanding?

  4. 4.

    How effective were your strategies in learning about motivation in this video?

  5. 5.

    What could you have done differently while learning about motivation?

The software program Silverback was used to capture participants’ verbalization as they thought aloud while learning with the video. Additionally, Silverback provided keystrokes and video capture for each participant.

2.2.2 Procedure

One of the two authors individually ran the participants. Following completion of the consent form, the participant was given an overview of the experimental session. The participant then completed the pre-test, which measured their prior knowledge of motivation theories. Next, the participant began learning with the video after the researcher read the directions.Footnote 1 The researcher remained nearby during the learning task to remind participants to keep verbalizing if they were silent for more then three seconds (e.g., “Say what you are thinking”). The participants’ verbalizations during the learning task were recorded and later used to analyze their SRL processes during learning. This approach, termed concurrent think-aloud, has an extensive history in cognitive psychology and cognitive science (see Ericsson 2006; Ericsson and Simon 1993; Newell and Simon 1972 for extensive reviews). Cognitive psychology and cognitive science have used both concurrent and retrospective think aloud protocols as data sources for cognitive processes (Anderson 1987). While the think aloud protocol has been most popular in reading comprehension, it has been shown as an excellent tool to gather verbal accounts of SRL and map out self-regulatory processes during learning (e.g., Azevedo 2014; Tuysuzoglu and Greene 2015). Verbalizing thoughts during learning will not disrupt the learning process, if the “…subjects verbalizing their thoughts while performing a task do not describe or explain what they are doing” (Ericsson and Simon 1993, p. xiii). If subjects are not asked to reflect, describe, and/or explain their thoughts during learning, but rather are asked to simply verbalize thoughts entering their attention, then it is assumed that the sequence of thoughts will not be disrupted. The experimental session lasted approximately 45 min for each participant.

2.2.3 Coding and Scoring

The raw data included approximately 800 min (13.33 h) of audio and video recordings from the 32 participants, as captured by the software program Silverback. In order to analyze the frequency of self-regulatory processes each participant verbalized during the learning task, the first author coded all of the Silverback recordings. This phase of data analysis yielded a total of 1135 coded SRL segments for all participants (M = 35.47 per participant). Inter-rater reliability was established by comparing the individual coding of the first author with the coding of the second author. The second author independently recoded 398 SRL segments (35 % of all codes). There was a strong positive agreement on the coded SRL segments (293 of the 398), as evidenced by the Pearson correlation (r = 0.74). Additionally, inter-rater reliability was established for the scoring of the pre-test and posttests by comparing the individual scoring of the first author with the scoring of the second author. The second author independently scored all 32 participants. Again, there was strong agreement between scorers (50 of the 64 pre/posttests,) as evidenced by the Pearson correlation (r = 0.78). All disagreements were resolved through the discussion.

3 Results

The first research question examines the relationship between metacognitive processes and the extent to which participants paused the video. Results from a Shapiro–Wilk test indicated that the pause variable distribution (p = 0.020) and the monitoring variable distribution (p < 0.01) deviate significantly from a normal distribution. As a result, a nonparametric test was used to test this first research question. A Spearman’s rank-order correlation was run to determine the relationship between the frequency with which the participants paused the video and the extent to which they monitored their learning. There was a strong positive correlation between pausing the video and Monitoring Understanding Plus (r s  = 0.46 p = 0.008) and Monitoring Understanding Minus (r s  = 0.59, p < 0.001).

A Mann–Whitney test was used to examine the second research question, which considered the differences between use of self-regulatory processes and mental effort between experimental conditions. Results indicate that the use of SRL processes was greater for those participants who received embedded prompts (Mdn = 47.18) than those participants who did not receive these prompts (Mdn = 23.69), U = 36.00, p = 0.001, r = 0.61. In order to analyze instructional efficiency for this question, a formula was used to calculate the difference between standardized measures of performance and learner effort (i.e. mental effort reported after the learning task; Paas et al. 2003). The difference between these two variables is divided by the square root of two. The resulting variable is a measure of instructional efficiency. Results indicate the condition with the embedded prompts was more efficient (Mdn = −0.42) than the condition without these prompts (Mdn = 0.42), U = 67.50, p = 0.022, r = 0.40. (see Table 1 for mean and standard deviations SRL and mental effort, by condition).

Table 1 Mean and standard deviations SRL and mental effort, by condition
Full size table

A repeated measures ANOVA was conducted to address the third research: To what extent do embedded prompts affect learning outcomes? Results indicate a significant effect of condition on learning outcomes, Wilk’s Lambda = 0.704, F(1, 30) = 12.589, p = 0.001, partial η2 = 0.30. Participants who received embedded SRL prompts while watching the video learned significantly more than those participants who did not receive these prompts (see Table 2 for mean and standard deviations for pretest and posttest, by condition).

Table 2 Mean and standard deviations for pretest and posttest scores, by condition
Full size table

4 Discussion

Support of flipped classrooms is gaining momentum as evidenced by its increasing prevalence in K-12 classrooms and higher education. This growing popularity is in part due to technological advances, which have made creating high-quality videos even easier. However, benefits may only be realized if students take advantage of this individualized delivery of instruction offered in flipped classrooms. Do students pause and replay the video when they are struggling with the content? What processes lead students to pause and replay the video? Can embedded prompts support SRL processes that facilitate learning with videos designed for flipped classrooms? These questions require research that collects process data, a methodological approach that was assumed by this study. Results indicated that participants’ monitoring of their understanding was significantly and positively related to pausing and replaying the video. However, the results also reveal that, on average, participants in the control condition rarely (if ever) monitored their learning, paused and replayed the video less frequently, and performed worse on the posttest. Furthermore, the significant difference in instructional efficiency between the two conditions suggests a positive “payoff” for embedding SRL scaffolds (van Gog et al. 2008). While the embedded SRL prompts positively affected both learning processes and outcomes, they did not negatively affect perceived mental effort. This study potentially offers implications for the design of flipped classrooms by highlighting the cost-effective approach of embedding SRL prompts in videos.

4.1 Design Implications

The emergence of the flipped classroom model and accompanying support by widely recognized organizations (e.g., EDUCAUSE) has produced a fairly robust body of work that describes how to most effectively create flipped classrooms. Despite this increasing popularity, research has found fairly mixed results on their effectiveness. These inconclusive results hold true even for undergraduate students. For example, Hashmi and Shih (2013) found that adult students are not adequately prepared to assume a higher level of responsibility of learning that is necessitated by the flipped classroom model. These findings suggest that flipped classroom designs need to include prompts that actively engage students while learning from videos. Students need to take an active role in learning with a video in order for the flipped model to be most effective. Results from this study support these assumptions; the participants from the control condition rarely engaged in self-regulatory behavior. However, results also indicate the inclusion of SRL prompts in videos provides the necessary support for active engagement in the learning process. It would seem that these undergraduates have the capacity to self-regulate their learning, but may require prompts to engage in self-regulation. Furthermore, results also indicate that this addition to the video did not degrade instructional efficiency, and in fact seemed to improve this important design component.

The concept of instructional efficiency is derived from the Cognitive Load Theory (CLT). Salomon (1984, p. 648) described cognitive load as mental effort representing, “the number of non-automatic elaborations necessary to solve a problem.” Cognitive overload can occur if the number of non-automatic elaborations processed exceeds the limited capacity of working memory. Three types of cognitive load (intrinsic cognitive load, extraneous cognitive load, and germane cognitive load) all have the capacity to consume the limited space of working memory (e.g., Gerjets et al. 2004; Kester et al. 2005; Sweller 2004; van Merriënboer and Sweller 2005; Verhoeven et al. 2009). Intrinsic cognitive load results from an interaction between a student’s level of prior domain knowledge and the inherent complexity of the content to be learned. The other two types of cognitive load are of particular concern to the instructional design of videos for flipped classrooms. Extraneous Cognitive Load results from design features that necessitate additional and unproductive processing for the student. Germane Cognitive Load, on the other hand, reflects processing that leads to increases schema construction and automation. Thus, a goal of the instructional design is to decrease extraneous cognitive load while enhancing learning outcomes. Assessing instructional efficiency provides a quantitative evaluation of instructional design in terms of cognitive load and performance. A potential concern of adding prompts to instructional videos designed is that the student will experience additional, unproductive mental effort in response to the prompts. Results from this study suggest indicate that embedded SRL prompts enhance learning outcomes while having minimal negative impact on mental effort. In other words, adding SRL prompts to these videos enhances instructional efficiency, a design implication that is important to consider as the flipped classroom model becomes increasingly commonplace.

4.2 Methodological Implications

Research is emerging that offers insight on the effectiveness of flipped classroom models, though much of this research has focused on students’ perceptions of this pedagogical approach (Hashmi and Shih 2013) or learning outcomes within this model (Davies et al. 2013). While this research has furthered our understanding, the field will be pushed forward with research that examines how students self-regulate their learning in flipped classrooms. In order to pursue this line of research, current SRL methodologies should be considered. SRL methodologies typically reflect assumptions regarding one of two properties, aptitude or event. An aptitude is a relatively enduring trait that can be used to predict future behavior. Based on this assumption, self-perceptions of self-regulation are considered valid measures of SRL. These perceptions often are derived from responses to questionnaires, with self-report questionnaires being the most frequently used protocol for measuring SRL as an aptitude (Winne and Perry 2000). Several self-report questionnaires are used most frequently, and include such self-report questionnaires as the Learning and Study Strategies Inventory (LASSI; Weinstein 1987) and the Motivated Strategies for Learning Questionnaire (MSLQ; Pintrich et al. 1991). Despite the ease of administration, self-report questionnaires have come into question regarding their alignment with core assumptions of SRL.

More recently, the field has advocated using protocols that assume SRL is a dynamic unfolding event and thus should be examined in real time. As such, recent research has advocated that SRL should be considered an event and data should be collected during learning. The think-aloud has emerged a predominant approach in capturing how students self-regulate during learning. The think aloud has an extensive history in cognitive psychology and cognitive science (see Ericsson 2006; Ericsson and Simon 1993; Newell and Simon 1972 for extensive reviews). As research continues to examine the effectiveness of flipped classrooms, it is important that research utilize methodologies that produce process data, such as the concurrent think-aloud, to better understand how students learn in this classroom model. Recent technological advances have created environments that offer additional methodologies to capture SRL in real time. For example, gStudy (e.g., Perry and Winne 2006) offers very fine-grained, stamped data while students learn with multimedia topics structured into packages called learning kits. MetaTutor offers another hypermedia learning environment that is based on extensive research by Azevedo and colleagues. This learning environment was designed to not only model and foster self-regulation when learning about human body systems (Azevedo 2008; Azevedo et al. 2009), but it also has the capacity to detect and trace self-regulatory processes through multi-methods. These researchers are producing cutting edge learning environment that integrate several technologies (i.e. adaptive embedded pedagogical agents with hypermedia), while also offering advanced methodological tools to capture process data. Capturing process data on SRL is not limited to these advanced technologies, though. The software used for this study (Silverback), for example, offers simplicity and an affordable price point. Utilizing the built-in iSight webcam and microphone, Silverback creates a video and audio file of the learning episode. Unlike gStudy and Metatutor, however, log files are not concurrently created. Rather, the audio and video recordings need to be separately coded. Given its simplicity and accessibility, however, Silverback offers an attractive methodological option for capturing SRL process data.

4.3 Future Directions

The complexity underlying the relationship between SRL phases and learning with videos was not fully examined in this study, an issue that can be more fully explored with future research. A fundamental assumption of the theory guiding this study is that SRL is comprised of three interactive phases: forethought, performance, and self-reflection. Analysis of the learning task and subsequent goal setting comprises the forethought phase. Motivational orientations play an integral role in this phase and set the stage for performance stage. The use of strategies during this phase, such as taking notes, metacognitive processes, and attention focusing strategies, facilitate learning. The final phase of SRL, self-reflection, occurs when students judge and develop reasons for their performance. Self-evaluations in this phase tend to be idiosyncratic because they reflect both the assessment criteria and the student’s individual performance level. These self-reflections reflect individualized goals established during the forethought phase, a relationship that highlights the dynamic and cyclical nature between the three SRL phases. Furthermore, students can experience negative or positive emotions when justifying their success or failure depending on their attributional style. In turn, emotions experienced during the self-reflection of SRL will influence future motivation and regulation. These assumptions suggest that phases interact dynamically and cyclically, which raise an interesting theoretical question: Is it necessary to prompt all SRL phases? The theoretical assumptions suggest that prompting one phase of SRL will affect other phases due to their interactive nature. Future research should consider the potential effect of prompting a single SRL phase, as opposed to all three phases as was done in this study. Prompting a single SRL phase may have differential effects depending on level of prior domain knowledge, motivation orientations, and/or developmental group. Additionally, this study focused on one component of a flipped classroom: videos designed to be used as homework. While findings from this study offer implications for the design of these videos, the classroom component of this model was not examined. An important aspect of this model is the ability to restructure interactions during class time and promote deeper learning. Future research would be well served to examine the relationship between redesigned videos, such as the one used in this study, and learning during class time.

4.4 Limitations

While this study offers implications for the design of flipped classroom videos, the limitations need to be addressed. First, the interpretation of the data should constrained to the developmental group of this study, undergraduate students. A robust body of research has demonstrated developmental differences with respect to SRL capacity (Pintrich and Zusho 2002). Older students tend to self-regulate their learning in a qualitatively different manner than younger students. For example, Vukman (2005) examined SRL processes (metacognition) with students from various developmental groups. This study found that the transition from adolescence to adulthood was marked by significantly more accurate reflection on one’s abilities, findings that have been replicated by many other studies. Younger students, who have qualitatively different SRL capacities, may distinctly respond to SRL prompts. While research has demonstrated that younger students have the capacity to self-regulate (see Nancy Perry’s work, for example), their expertise level and other maturational factors may moderate the effectiveness of SRL prompts.

Second, the relatively small sample size of this study needs to be acknowledged. The time intensive nature of the methodology and data analysis necessitated this sample size. Participants were individually run through the experimental session, and think-aloud data from this experimental session was transcribed, coded, and then re-coded for inter-rater reliability. While this time intensive collection process yielded rich SRL data, it limited the number of participants the researchers were able to run through the study.

4.5 Conclusion

The flipped classroom model has been gaining substantial popularity, in part due to the opportunity offered to students. Content delivery in flipped classrooms is more highly individualized because videos assigned as homework provide students with the opportunity to control the pacing of information. Furthermore, this model allows teachers to reallocate precious class time and provide more opportunities for application activities, class discussions, and individual attention. Despite these tantalizing promises, though, research has found mixed results on flipped classrooms. This study contributes to our understanding of flipped classrooms by providing process data on how students learn with videos designed for this model. Results from this study indicate that monitoring of understanding is significantly related to pausing a video during learning. Additionally, participants who received embedded prompts while learning with the video used more self-regulatory processes. Given the positive effect of prompts on SRL, it was not surprising to find that participants who received these prompts performed significantly better on the posttest. Furthermore, the addition of these prompts did not degrade instructional efficiency, suggesting that there is little “cost” to their inclusion. In sum, this study contributes to the field by offering methodological and design implications for the flipped classroom model.