Effectiveness of the flipped classroom on self-efficacy among students: A meta-analysis

Abstract Increasing attention has recently been paid to the flipped classroom (FC) due to advancements in information technology. Self-efficacy plays a crucial role in the learning effect of students. However, the impact of flipped classrooms on students’ self-efficacy remains uncertain. In this study, we have developed a theoretical framework focusing on self-efficacy to understand its relationship with the flipped classroom better. This meta-analysis investigates the influence of the FC approach on self-efficacy across various educational settings. A total of 22 articles were selected in the meta-analysis, revealing that the FC approach can enhance self-efficacy, thereby potentially increasing student engagement in learning. Subgroup analyses revealed that technicality subjects and the short duration of the intervention fit the FC more. These results emphasize the significance of considering self-efficacy in the context of flipped classrooms and highlight the approach’s benefits. Furthermore, careful course design is essential to optimize the effectiveness of flipped classrooms. However, due to the heterogeneity and limitations of the included studies, the results should be interpreted with caution. Future research is recommended to explore further the effectiveness of the FC approach on self-efficacy in various educational settings.


Introduction
In recent years, there has been a growing interest in the flipped classroom (FC), which is gaining popularity due to the rapid advancement of modern information technology.It is widely used in various countries and regarded as an innovative teaching strategy (Chu et al., 2019).

Advantages of the flipped classroom
There are several differences between flipped classrooms (FC) and conventional lecture-based learning (LBL).LBL traditionally follows the teacher-centered approach where students passively absorb information during the teaching process, and the teacher schedules lectures (Freeman et al., 2014;Theobald et al., 2020).On the other hand, the FC method, built on constructivism, is a student-centered approach that motivates students to increase their grasp of and familiarity with the knowledge system through active engagement and reflection (Piaget, 2003).Hwang and colleagues propose some critical characteristics of the FC, including (1) changes in the use of inclass time (lecture in LBL) and out-of-class time (homework in LBL) and (2) changes in the learning status of students during in-class and out-of-class (Hwang et al., 2015).According to Blanco et al. (2016), students will use videos, textbooks, or learning platforms offered by teachers to learn the material previously taught by teachers in class.The FC strongly emphasizes peer and teacher interaction in the classroom, shifts the focus from lecturing to problem-solving, and transforms students from passive consumers of knowledge to active participants in their learning process.Recent research reveals active learning helps students understand concepts and perform better (Chen et al., 2018).
Since the FC was brought up, there has been much interest in comparing the FC to the LBL.Day (2018) compared FC with conventional gross anatomy lectures and provided evidence of FC's contribution to raising academic performance.According to research by Zhao et al. (2021), who used a task-driven instructional strategy in the FC, students who received flipped learning teaching scored higher on the final test than those in the LBL.Similarly, Łopińska et al. (2022)discovered that students receiving FC outperformed those receiving regular LBL regarding test scores.Based on several studies that have been published, students favor FC over conventional LBL (Gallardo et al., 2022;Joseph et al., 2021;Wu et al., 2022).
However, not all studies on flipped learning have consistently beneficial results.Smith et al. (2020) hypothesized that students' test scores were marginally lower in FC settings compared to LBL settings.Similarly, Wilson and Hobbs (2022) suggested that no significant difference in academic achievement was found between the FC and the LBL environments.Researchers have conducted meta-analyses to synthesize findings from various individual investigations in the context of flipped classrooms (Cheng et al., 2018;Shi et al., 2019;Van Alten et al., 2019).Specifically, Shi et al. (2019) gathered 33 pertinent studies contrasting FCs with LBLs and conducted effect size assessments.According to the findings, FCs benefit students' cognitive learning.However, another meta-analysis suggests that the effectiveness of FCs in enhancing cognitive learning is somewhat limited (Cheng et al., 2018).A meta-analysis of 19 studies of flipped classrooms meta-analysis was done in another study, and the results showed that the influence on students' cognitive learning was favorable (Hew et al., 2021).It is also said that these results should be interpreted cautiously due to the substantial heterogeneity of the meta-analyses caused by factors including course design in primary studies (Hew et al., 2021).Although previous research has generally supported the benefits of flipped classrooms, it is essential to recognize the potential problems.

A factor in the flipped classroom: student engagement
Numerous studies have looked into the factors that influence the effectiveness of FC (Hew & Lo, 2018;Hu et al., 2018;Shi et al., 2019).The factors can be broadly categorized into two types.The first category is devoted to teacher-controlled factors, where the teacher's efforts influence the FC result in curriculum design and quality management (Cheng et al., 2018).The other category comprises student-controlled factors, requiring them to actively participate in pre-class activities to perform better on the FC (Diwanji et al., 2018).Previous research consistently indicates that high levels of student engagement improved student performance in FC contexts (Cheng et al., 2021;Han, 2022;Lai et al., 2021).Reeve (2012) defines student engagement as characterized by active participation in academic, extracurricular, and school-related activities and dedication to learning objectives.Blumenfeld and Paris (2004) further break down student engagement into three components, behavioral, emotional, and cognitive, as described.Behavioral engagement is a dimension related to students' effort, including their participation in coursework, execution of tasks, and attendance.Emotional engagement is connected to their emotional reactions, such as how satisfied they are with the course and how much stress they feel.Cognitive engagement relates to students' cognitive processes, self-regulation, and self-efficacy.A recent cross-sectional study revealed a positive correlation between students' academic achievement in LBL and their cognitive engagement (Dogan, 2015).Lo and Hew (2021) conducted a meta-analysis that incorporated numerous studies on FCs and examined student engagement in three dimensions.They discovered that FCs can improve students' behavioral engagement and emotional engagement.Improvements in the student's comprehension of the course material and their preference for complex tasks were seen in terms of cognitive engagement.However, the outcomes for self-efficacy were mixed.Some research in the article revealed that students who engaged in FCs had stronger selfefficacy than students who participated in LBLs.In comparison, some students in FCs exhibited poorer self-efficacy than those in LBLs (Kennedy et al., 2015).Other research found no difference in self-efficacy between the two teaching contexts (Wasserman et al., 2017).It is important to note that these studies did not compare the pre-and post-intervention changes in FC intervention, but focused on the differences between cross-sectional FCs and LBLs.Additionally, the study included a relatively limited number of related studies, which could introduce potential biases in the results.In light of this, exploring self-efficacy within the context of student engagement remains a significant area of study.

Self-efficacy to achievement: a virtuous cycle
Bandura defines self-efficacy as "people's assessments of their capacities to plan and carry out the courses of action necessary to achieve specified performances (Bandura, 1986)."In other words, it relates to a person's self-perception of their ability to act.Since Bandura first established the concept of self-efficacy, it soon gained popularity and was extensively embraced in education.Self-efficacy in education refers to assurance in one's capacity to learn or behave in a particular way (Schunk & Pajares, 2001).It is both a predictor of academic success and an indicator of students' psychological health (Meng & Zhang, 2023).Many research investigations have shown a link between academic performance and self-efficacy (Paciello et al., 2016), with self-efficacy being the best predictor of academic success (Dogan, 2015).In particular (Schunk & Pajares, 2010), observed that individuals with higher levels of self-efficacy frequently exhibit greater intrinsic motivation, set more challenging goals, and uphold high levels of commitment.Moreover, selfefficacy is widely recognized as essential to motivation, significantly impacting student engagement (Bond, 2020).One research proposed that the positive relationship between autonomous motivation and behavioral engagement increases when self-efficacy increases (Lai et al., 2021).Additionally, when self-efficacy is high, students are more likely to engage in class activities as they believe they have the knowledge and abilities to handle challenges (Joshi & Kuhn, 2011).In contrast, individuals with poor self-efficacy may experience anxiety when faced with unfamiliar learning situations (Jolie & Lee, 2006).Students with higher levels of self-efficacy typically outperform their peers academically, while those with lower self-efficacy may struggle to succeed in school to the same degree.
Furthermore, studies have demonstrated that self-efficacy can predict student participation (Azila-Gbettor et al., 2021;Ferrell, 2012;Singh & Abdullah, 2020).Students with higher levels of self-efficacy are more likely to participate in class and demonstrate better levels of active learning.A study with 1158 participants showed that self-efficacy directly predicts academic success, with student engagement mediating between the two (Luo et al., 2023).Another study with 258 college students supported this finding (Meng & Zhang, 2023).Likewise, multiple lines of research indicate that students with high levels of self-efficacy think they are more involved in the FC (Palazón-Herrera & Soria-Vílchez, 2021;Sun et al., 2018).Meanwhile, A general self-efficacy model put forth by (Linnenbrink & Pintrich, 2003) contends that self-efficacy raises levels of engagement (such as behavioral, cognitive, and motivational engagement), boosting learning and achievement.Over time, the result of achievement flows back to self-efficacy and creates a virtuous cycle.Self-efficacy can generally increase student engagement and improve instruction quality in the flipped classroom.Therefore, raising students' self-efficacy is crucial for academic success.
Drawing upon previous models (Chiu, 2022;Lai et al., 2021;Linnenbrink & Pintrich, 2003;Reeve, 2013) that elucidate the relationship among self-efficacy, student engagement, and achievement, we have developed a comprehensive theoretical framework (Figure 1) that emphasizes the role of self-efficacy in enhancing the effectiveness of flipped classrooms.In this theoretical framework, self-efficacy serves as both the initial catalyst and the ultimate outcome.Specifically, Self-efficacy plays a pivotal role in the dynamics of flipped classrooms.It fosters students' engagement, which, in turn, positively influences their academic and psychological outcomes within the FC environment.As students succeed, their self-efficacy is further reinforced, creating a virtuous cycle of continuous improvement.In light of this, it raises the question of whether students with lower initial levels of self-efficacy can also benefit from this transformative process.
While numerous meta-analyses have been conducted on FCs, to our knowledge, none have focused explicitly on self-efficacy in the context of FCs.Therefore, our study seeks to bridge this gap by investigating the potential of FCs to enhance student self-efficacy.We aim to offer insightful information to aid in advancing FC methodologies.To achieve this, we conducted a meta-analysis not confined to a single discipline or country to address the following research questions (RQs).
RQ1: Does the FC approach result in higher levels of self-efficacy compared to the LBL approach?RQ2: Does implementing the FC increase students' self-efficacy as measured by pre-and postintervention measures?

Methods
This meta-analysis followed the guidance of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analysis) statement (Page et al., 2021).Meta-analyses must be registered on PROSPERO by PRISMA criteria to prevent job duplication.We have the PROSPERO registered ID CRD42023394644.The data used in this meta-analysis were all from published studies.Therefore, ethical support and student consent were unnecessary.

Conceptual framework
Student self-efficacy assesses their capacity to carry out learning tasks in a particular topic.In the words of Bandura et al. (1997), this judgment significantly impacts people's motivational styles and behavioral patterns, including how much effort they put forth and how they handle challenges.In this study, we examined the impact of FCs on self-efficacy from both cross-sectional and longitudinal perspectives.We intended to explore how student self-efficacy varied across flipped and traditional classrooms and how it changed due to FC interventions.
The FC is a student-centered learning method that emphasizes students' independent learning more than teachers' direct instruction (Galindo, 2014;Lage et al., 2000).In other words, activities previously done outside the classroom are being done during class time, while conventional inclass activities are shifting outside the classroom.This article describes the FC method as any preclass activity that entails learning classroom content, regardless of the particular instructional design.
Research has found that when contrasting FC with LBL, the effectiveness of the FC approach can be affected by different types of instructional designs.Based on Bandura's social cognitive theory, we understand that human behavior is determined by the interaction of individual cognition and the external environment in which individuals live (Bandura, 2012a).By applying this idea to our study, we can observe that FCs are also influenced by the interaction of student characteristics (individual factors) and classroom settings (environmental factors).As a result, we will separate these variables into three subgroups for our forthcoming analysis: (1) age, (2) academic subject, and (3) intervention duration.Age represents the individual factor of students, while academic subject and intervention duration represent environmental factors.
According to prior studies, FCs benefit individuals of all ages (Gargallo-Camarillas & Teruel Ferrer, 2021).Yet this study discovered that older students are more self-directed learners than younger students, suggesting that FCs perform better for older students.Additionally, various age groups are frequently linked to various educational stages.For instance, teenagers between 14 and 18 often enroll in high school, but adults over 25 May be employed or pursuing graduate degrees.These various educational phases could require various curriculum designs and methods of instruction.
Due to their distinctive characteristics, different academic subjects may call for various approaches to course design.Results from earlier meta-analyses evaluating effect sizes across fields have been inconsistent.Results from (Cheng et al., 2018) and (Låg & Saele, 2019) showed significant differences in effect sizes between subjects, with humanities subjects having a more significant effect size than STEM subjects.However, neither (Van Alten et al., 2019) nor (Shi et al., 2019) discovered significant differences between the subjects.Our study is interdisciplinary; thus, we will count that as one of the factors affecting the outcomes.
Although earlier studies have found no significant differences in study or implementation duration (Cheng et al., 2018;Shi et al., 2019), our analysis includes single-session FCs and FCs extending for a few weeks or more.These two kinds of interventions take noticeably different amounts of time.We will, therefore, use it as a variable in the subgroup analysis.The conceptual framework of our investigation is depicted in Figure 2.

Search strategies
The search was carried out in January 2023, and three online databases were used: PubMed, Science Direct, and Web of Science.Literature was collected from the point of database establishment to 26 January 2023.Databases were used to search key terms including "flip*", "class*", and "self-efficacy".The asterisk (*) is a wildcard symbol used to widen the scope of material included and obtain conceptually similar content.This search strategy was conducted independently using three online databases.

Eligibility criteria
A meta-analysis of literature that matched the eligibility criteria was required to examine the effectiveness of self-efficacy in the FC.The eligibility criteria are listed in Table 1 and include the following: (a) studies conducted a comparative analysis, comparing FC and LBL, or utilizing a within-subject design with pre-and post-intervention comparisons for the same group of participants; (b) a self-efficacy survey and scores, for which "confidence" has the same meaning as "self-efficacy"; thus, scores of "confidence" can also be included; and (c) the results can be used for quantitative analysis.Duplicate reporting was eliminated.org/online-learning/core-software/revman).The two reviewers independently gathered data from the comparative studies.The reviewers evaluated, discussed, and resolved inconsistencies between the extracted data before data input and analysis.

Assessment of methodological quality
The Medical Education Research Study Quality Instrument (MERSQI) assessed methodological quality (Reed et al., 2007).This approach has been employed in other meta-analyses of flipped classrooms (Hew & Lo, 2018).Since nearly half of the studies in this research are related to medical education, this method was employed to assess the methodological quality.It includes ten items,  reflecting six domains in study quality: (a) study design, (b) sample, (c) type of data, (d) validity, (e) date analysis, and (f) outcomes.For each domain, the maximum score is 3. Thus, the maximum score of MERSQI is 18, and its potential range is from 5 to 18 (Smith & Learman, 2017;Smyer, 2011).

Data analysis
The meta-analysis was performed using RevMan (Version 5.4).To compare the effect size, we used a random effect model.Because the self-efficacy score was likely affected by different conditions, we used the mean and standardized difference (SD) of the self-efficacy score extracted from the literature to compute the standardized mean difference (SMD) as the effect size.The difference was statistically significant when the two-tailed P value was < .05.The I 2 -statistic was used to express the level of heterogeneity (i.e., inconsistency level attributed to study outcome).
A subgroup analysis was conducted to explore the effect of different contexts on the outcome measure and identify differences among subgroups.Publication bias was assessed by Egger's test and funnel plot.

Results
As can be seen from Figure 3, the literature search process is depicted by a PRISMA flowchart.A total of 2800 articles were included in the initial search.However, we found that many articles lacked inclusion criteria after screening the title and abstract, and information related to a selfefficacy scale or empirical research was absent.For instance, a host of irrelevant articles from the collected literature failed to gather and analyze any data that involved, for example, inverted repeats, flipped source followers, and triple spin flips).Finally, 67 full texts were read and assessed for eligibility.Of these 67 articles, 19 were removed because they did not incorporate a selfefficacy scale, and a further 11 were eliminated due to insufficient data on students' self-efficacy.
The final result was that 22 papers were considered, with sample sizes ranging from 15 to 367 individuals, and the entire sample size was 2353.
Activities in the FC are summarized in Table 3.The outcome indicators included self-efficacy and confidence scores.

Quality assessment
The methodological quality assessment included studies detailed in Table 4.The mean (SD) MERSQI score in this meta-analysis was 10.2 (0.68), representing the potential range of 5 to 18.

Meta-analysis
A thematic analysis method was used in the meta-analysis to answer the RQ offered in this study.
To address RQ1, a comparison between the FC and the LBL setting was conducted to investigate if students in the FC showed higher levels of self-efficacy during the intervention than those in the LBL setting.To address RQ2, a pre-and post-intervention study was conducted to investigate whether students' self-efficacy increased due to participating in the FC.Additionally, subgroup analyses based on age, subject, and intervention duration were carried out to assess the heterogeneity in these two analyses.

Subgroup analysis
To explore the possible source of heterogeneity, we analyzed variables using the random effect model.Figures 5, 6 , and 7 detail the subgroup analysis results.There was still significant heterogeneity among the various groups in the age subgroup (Figure 5).Age was not the cause of heterogeneity.Furthermore, no significant difference between the groups was found according to the test for subgroup differences.(P = 0.32, I 2 = 11.7%).

Study
Significant differences for subjects were discovered in this subgroup (Figure 6).The results of the subgroup analysis by subject showed that this was not the cause of heterogeneity.The test for subgroup differences indicated significant variation between the groups (P = 0.009, I 2 = 78.6%).This implies that using FC approaches to various subjects may have varied results.
For the duration of the intervention, a subgroup analysis was also conducted, revealing significant differences among the three subgroups (Figure 7), which illustrated that the duration of the intervention was not the source of heterogeneity.The test for subgroup differences indicated significant variation between the groups (P = <0.001,I 2 = 86.2%).This suggests that depending on the duration of the intervention, the outcomes of applying FC methods may change.

Publication bias assessment
As can be seen from the funnel plot, the self-efficacy score was slightly asymmetrical, and a low likelihood of publishing bias was found (Figure 8).

Subgroup analysis
The results of the subgroup analysis are illustrated in Figures 10, 11 , and 12.
We found no significant differences in the age subgroup, indicating no variation among the age subgroups (P = 0.85, I 2 = 0%, Figure 10).Furthermore, each age group exhibits significant heterogeneity, which suggests that the age subgroup is not the source of heterogeneity in the metaanalysis.
Significant differences were observed among the three subgroups in terms of subject (P = 0.02, I 2 = 74.4%, Figure 11), suggesting that the subject of the course may impact the FC's effectiveness.However, due to the high heterogeneity within each subgroup, the subgroup analysis by subject did not identify a specific source of heterogeneity.We conducted a subgroup analysis based on the duration of the intervention.Firstly, a significant difference was observed among the subgroups (P = 0.002, I 2 = 79.5%, Figure 12), indicating that the duration of the FC intervention may influence its effectiveness.In the heterogeneity analysis, no significant differences were found in the "less than 1-week" subgroup (P = 0.18, I 2 = 39%), while significant differences were observed in the other two subgroups, suggesting that the duration of the intervention was not the source of heterogeneity.We discovered that FCs lasting less than a week had more substantial effects and higher internal consistency, suggesting that the FC approach might be more appropriate for short-term instruction.

Publication bias assessment
As can be seen from the funnel plot, the self-efficacy score was relatively symmetrical, combined with the value of Egger's test (p = 0.082), which was higher than 0.05, and no publishing bias was found (Figure 13).

Discussion
Overall, we conducted a meta-analysis of 22 studies comparing self-efficacy differences between the FC and LBL settings and changes in pre-intervention and post-intervention surveys.First, the study's statistics show that students in the FC setting have a bit greater self-efficacy than LBL students.Second, there was statistically significant evidence of further development in students' self-efficacy in the FC when pre-and post-intervention results were compared.The FC, in our opinion, positively affects students' self-efficacy overall.

Factors influencing the flipped classroom
We found significant subject and intervention duration differences in the two subtheme analyses.This finding may imply that various courses call for different instructional designs and that certain subjects may not be appropriate for FC instruction.This finding is consistent with previous research (Cheng et al., 2018).In our study, across FC courses, Nursing demonstrated the most growth in  self-efficacy, whereas Mathematics demonstrated the most negligible growth and even a negative effect.We cannot ignore possible issues with self-efficacy even though some research has shown successful implementation of FCs in mathematics (e.g.Sun et al. (2018)).
Given the duration of the intervention, students' self-efficacy improved more when the flipped classroom was implemented for shorter periods.This can be due to the novelty impact during the early stages of implementing FC when students might be interested and engaged.However, as time passes, they may grow weary or bored or get anxious from the increased burden compared to  LBLs (Dong et al., 2021).This may also depend on how the pre-class activities are structured.Future studies might concentrate on specific subjects or methods within the FC.For instance, examining the FC's effects in a specific academic subject matter, like the humanities or mathematics, may provide more focused insights into how it affects self-efficacy.The best methods for boosting self-efficacy may also be determined by researching various FC variations, such as using internet platforms, interactive films, or collaborative learning.Future research can advance our understanding of the FC's usefulness in building self-efficacy in various educational environments by moving deeper into these particular elements.
Interestingly, in addition to enhancing self-efficacy in general settings, a study conducted by Croy et al. (2020) explored changes in self-efficacy in stressful situations.The findings indicated that self-efficacy remained stable between the pre-and post-tests.By fostering students' selfefficacy traits such as confidence, capacity, tenacity, and strength, the flipped classroom strategy effectively mitigated the anxiety-inducing nature of the situation (Conner, 2015).In future research, further exploration can be conducted to investigate the effectiveness of flipped classroom approaches in different contexts or situations.
However, some studies have shown that the FC cannot increase self-efficacy (Kennedy et al., 2015;Tsai et al., 2022).This may be because the transition between the FC and the LBL during the studied period was too abrupt, and students may have needed more time to mentally and behaviorally accept the new teaching method.Previous research found that some students did not engage in learning during the pre-class phase, which meant they could not keep up with their educator during the in-class phase, so they gave up learning (Herreid & Schiller, 2013).Furthermore, some students have disliked collaborative group work (Awidi & Paynter, 2019).Another possibility is that the FC provides too much content (Dong et al., 2021).It causes students to feel too much pressure, which leads to anxiety.Research has demonstrated a negative relationship between anxiety and self-efficacy (Jolie & Lee, 2006).From the above, adopting a rational attitude when designing a course is crucial to ensure students can accept the new method.

Implications for future flipped classroom design
Based on previous research and Figure 1 findings, it can be inferred that raising self-efficacy can help FCs achieve better results, while raising FC effectiveness can also raise personal self-efficacy.Both of these components may combine to create a virtuous cycle.Self-efficacy theory can be incorporated to improve FC design further.
Enactive mastery experiences refer to actively acquiring experiences of success or failure (Bandura, 2012b;Bandura et al., 1997).Enactive mastery experiences are used in education to boost students' positive self-perceptions by drawing on prior accomplishments.Enactive mastery experiences have been recognized as the primary source of self-efficacy (Zhang & Ardasheva, 2019).Students must succeed to obtain enactive mastery experiences as they are built upon prior experiences.Students may experience setbacks when faced with the unknown, but they will feel accomplished when they use their talents to solve difficulties.Teachers can gradually increase the complexity of the material in their courses so that students can experience success before moving on to the next level.This reduces students' fear of failing and helps them become more confident in the subject.
Vicarious experiences stem from observing successful social models (Bandura et al., 1997).Specifically, vicarious experiences involve individuals observing their own or others' successes and failures (Schunk & Hanson, 1985).Students inspire desire and belief in their abilities when they see other students succeed or receive rewards due to their efforts.Collaborative group work and whole-class questioning can provide more opportunities for students to gain vicarious experiences (Wiggins et al., 2017).Verbal persuasion is observed when instructors clarify or affirm students' abilities, increasing their confidence.Teachers can use verbal persuasion to encourage students and help them believe in themselves.This belief in their abilities enhances their perseverance and equips them to overcome challenges.From the students' perspective, their self-efficacy improves when they receive information that supports their understanding (Naibert et al., 2020), increasing confidence in their ability to accomplish tasks.On the other hand, if teachers provide students with knowledge that challenges what they believe they already know, it could stimulate their attention and encourage a desire to learn (Shin & Kim, 2019).Such contact is valuable for education.Previous research has also pointed out the beneficial association between teacher feedback and self-efficacy (Turner et al., 2014).The necessity of face-to-face training in FC design is an issue of continuing discussion.However, current research shows that face-to-face components in FCs produce better results than those without them (Van Alten et al., 2019), and tests in the classroom increase the efficiency of FCs (Hew & Lo, 2018).These results emphasize the value of quick feedback.Unfavorable findings from a study on the flipped classroom approach's effectiveness point to a possible link between the lowered teaching effectiveness and the amount of face-to-face training (Gundlach et al., 2015).According to the self-determination theory, immediate feedback is essential for raising students' cognitive engagement and sense of self-efficacy (Niemiec & Ryan, 2009).Students also value teachers who use feedback to track their development and correct any difficulties they face (Skovholt, 2018).
Students' physiological and emotional states are triggered during the learning process, and if they respond positively to what they are learning, their self-efficacy is further enhanced.On the contrary, negative emotions, such as stress or anxiety, in a learning environment could be perceived as a sign of inability (Sawtelle et al., 2012).Therefore, it is essential to consistently monitor the physical and mental well-being of students throughout the learning process in the curriculum.
The design of the curriculum requires effort from teachers as well.The pre-class activities in this study's FC design provide the most statistical merging challenges.This is because there are so many kinds of pre-class activities that it is difficult to group them into a few separate groups.
Furthermore, previous research has indicated that different pre-class activities can affect FC outcomes (Hew & Lo, 2018).Following Bandura (1977)'s work, three dimensions of self-efficacy were proposed: magnitude, strength, and generality.The magnitude dimension describes the task's degree of difficulty.It is crucial to consider the magnitude of the dimension of self-efficacy while designing pre-class exercises to account for the degree of difficulty.The difficulty should not be highly demanding or complex to avoid student frustration and the decision to give up and not ultimately engage in the FC's learning process.
The zone of proximal development, which refers to the range of abilities that an individual can accomplish with the assistance of an expert but cannot yet perform independently (Ness, 2020), should be used to determine the complexity of pre-class exercises.Individuals' curiosity may be aroused by cognitive incongruity brought on by the failure to accomplish a goal (Shin & Kim, 2019), heightening their interest in this subject.The effectiveness of pre-class exercises is also vital since students prefer to feel confident in their ability to learn the subject before making an intentional effort (Ale Ebrahim et al., 2020), which increases their engagement and confidence.
In addition, evidence suggests that the increase in self-efficacy brought about by FC can have a long-lasting impact, even if our study revealed that a shorter time of FC was more successful in boosting individual self-efficacy.A follow-up study showed that students' self-efficacy, which increased during the FC, returned to baseline after 10 months (Bouwmeester et al., 2019).This may indicate that the FC can increase self-efficacy in the short term and that a more comprehensive long-term curriculum may be required to maintain higher levels of self-efficacy, as developing self-efficacy requires prior experience in overcoming challenges and difficult situations through continuous effort and perseverance (Bandura et al., 1997).
To sum up, maximizing students' self-efficacy enhances FC effectiveness.Based on the theoretical model depicted in Figure 1, using FC to enhance self-efficacy is the crucial initial step in facilitating students' advancement to the subsequent stages of learning within the FC setting.However, based on a meta-analysis, it is evident that FCs can be a feasible instructional strategy if adequately designed (Van Alten et al., 2019).Teachers, therefore, should carefully craft the FCs' content and develop their students' self-efficacy through a recurrent application.A positive selfefficacy cycle emerges in the context of FCs because students' self-efficacy motivates them to engage fully in the course and learn more.

Strengths and limitations
A positive relationship exists between self-efficacy and student engagement in learning, and increasing self-efficacy can significantly benefit student learning.As an emerging teaching strategy, the FC may enhance students' self-efficacy.To the best of our knowledge, this is the first meta-analysis to examine the effectiveness of the FC on self-efficacy.However, our meta-analysis had several limitations.First, some of the studies included in the current research were vague regarding their description of the course design, course length, course format, and course activities.Differences in these areas will affect the development of self-efficacy.
Secondly, our meta-analysis primarily focused on comparing the pre-and post-intervention effects of the FC and its comparison with the traditional LBL setting.However, it is worth noting that emerging teaching strategies, such as problem-based learning, peer-assisted learning, and gamified FC, have shown the potential to enhance students' self-efficacy.For example, Kim demonstrated the effectiveness of problem-based learning in improving students' self-efficacy (Kim, 2022).Zou et al. (2021) developed a digital game to examine its effects on English language learning among college students, revealing a significant enhancement in self-efficacy through game-based learning.Future studies could further investigate these alternative approaches to determine their comparative effectiveness in enhancing individual self-efficacy compared to the FC.
Third, it should be noted that, in this meta-analysis, self-efficacy was specific to the subjects the students were learning.It is unclear whether enhancing course-specific self-efficacy contributes to increasing individual self-efficacy.
Finally, it should be noted that the number of available studies is relatively limited, and substantial heterogeneity exists among them.Therefore, caution should be exercised when interpreting the findings.

Conclusion
This meta-analysis showed that students increased their self-efficacy after participating in the FC.This finding suggests that the FC approach can lead to increased self-efficacy, which can raise student engagement in learning.Furthermore, we found that the subject matter and the duration of the whole FC flow influence its impact on self-efficacy.In future course design, it is advisable to select subjects that are well-suited for flipped classroom implementation.Additionally, a blended approach integrating FC and LBL can be adopted to optimize the advantages of the FC, promote a thorough comprehension of the subject matter, and mitigate any potential negative emotional responses to its implementation.

Figure
Figure 1.Framework diagram for the promotion of flipped classroom effectiveness through self-efficacy.
Two reviewers independently screened the literature by reading the title and abstract to avoid subjective bias.Subsequently, they utilized the previously established eligibility criteria to review the complete text, ensuring a thorough and unbiased selection process.Data extraction needed to extract the following information: (a) writer; (b) year of publication; (c) sample; (d) type of course; (e) type of student; (f) study design (e.g., randomized controlled trials); (g) duration of intervention; (h) details of the FC; and (i) outcomes.If a disagreement arose between two reviewers, it was resolved by consulting a third party.Two reviewers cross-checked the data before being entered into the Cochrane Review Manager software tool (RevMan Version 5.4 https://training.cochrane.

Figure 7 .
Figure 7. FC vs. LBL: subgroup analysis of duration of intervention indexes.

Figure
Figure 8. Funnel plot of the FC versus the LBL self-efficacy scores.SE = standard error, SMD = standardized mean difference.

Figure 9 .
Figure 9. Forest plot of pre-and post-intervention for FC: selfefficacy indexes.

Figure
Figure 11.Pre-and postintervention: subgroup analysis on the subject.

Figure
Figure 12.Pre-and postintervention: subgroup analysis of the duration of the intervention.

Table 1 . Inclusion and exclusion criteria Inclusion criteria Exclusion criteria
Have comparison (randomized or non-randomized, interventional) Research questions follow through the PICO (Population, Intervention, Comparison, Outcome) criteria: (P) Students (any discipline, any level) (I) Use the FC approach in any format (C) Traditional lecture-based learning or pre-and post-test (O) Scores on self-efficacy Outcomes can be used for quantitative analysis Written in the English language Reviews, qualitative studies Outcomes have not reported self-efficacy The control group is also the FC Duplicate reporting (studies from the same author or institution)

Table 3 . Activities in the FC activities count
.,