Flipped Classroom Teaching and ARCS Motivation Model: Impact on College Students’ Deep Learning

Abstract

This study examines the impact of combining Keller’s ARCS motivation theory with the flipped classroom teaching model on the deep learning of college students. Using data collected from 495 students across different regions in China, the study employs Partial Least Squares Structural Equation Modeling (PLS-SEM) to investigate the relationships between motivational factors and deep learning. The findings reveal that attention, relevance, confidence, and satisfaction all significantly influence deep learning. Although relevance directly enhances deep learning, its effect on intrinsic motivation is less pronounced. Furthermore, the study reveals a hierarchical relationship among the ARCS dimensions within the flipped classroom context: attention drives relevance, relevance enhances confidence, and confidence leads to satisfaction—collectively supporting a sustained learning process. These results validate the application of the ARCS model in flipped classrooms, highlighting its potential to stimulate critical thinking and improve cognitive engagement. This research contributes to the theoretical development of motivation-driven learning models. It offers practical strategies for educators to optimize instructional design, thereby fostering sustained intrinsic motivation and deep learning among students.

1. Introduction

The rapid development of globalization and information technology has led to an increasing demand for innovative, high-quality talent in society. In higher education, it is essential not only to impart professional knowledge to students but also to focus on cultivating their innovative abilities and critical thinking skills (Pithers & Soden, 2000). Deep learning, which centers on advanced thinking and intrinsic understanding, can effectively enhance knowledge transfer, problem-solving skills, and autonomous learning abilities (Svinicki, 2004). Research shows that deep learning is closely linked to the future professional development of university students (Zhang & Wu, 2024). Deep learning emphasizes not only the mastery of knowledge but also the understanding of conceptual meaning, the construction of knowledge structures, and the ability to apply knowledge in new contexts. It helps students shift from fragmented, surface-level learning to systematic, reflective learning, significantly enhancing their cognitive development and learning outcomes (Biggs, 1979). In higher education, promoting the implementation of deep learning has become a core goal for cultivating higher-order thinking skills. However, the implementation of deep learning is not automatic; it requires stable and sustained intrinsic motivation to support it.

However, the current lack of intrinsic motivation among students has become a key issue limiting the implementation of deep learning. Studies have found that many university students in China experience learning burnout (Liu et al., 2023), with a predominant reliance on external motivation, such as grades and scholarships, while intrinsic motivation remains weak (Próspero et al., 2012). Because external motivation is less stable and tends to dissipate when external factors are no longer present, it further exacerbates the decline in learning efficiency and depth. Additionally, the societal phenomenon of “involution”, the pressure of graduation and employment, and the increasing demands for higher-level talent development have placed university students under considerable pressure from academia, employment, family, and society, making it difficult for them to focus solely on knowledge acquisition. This reflects the inadequacies of the current educational model in stimulating students’ intrinsic learning motivation.

In the context of traditional teacher-centered teaching models, which overly emphasize the one-way transmission of knowledge, educational reforms have introduced various innovative teaching methods to promote students’ autonomous learning and deep learning abilities. Among these methods, flipped classrooms have attracted significant attention in recent years due to their disruptive characteristics, which challenge traditional teaching and learning roles. The core of the flipped classroom is to shift the knowledge delivery phase to outside the classroom, allowing students to learn basic concepts through videos or online resources, while the in-class time focuses on discussions, practical activities, and the deeper application of knowledge (Bergmann, 2012). This model provides a more favorable environment for stimulating students’ intrinsic learning motivation and fostering deep learning strategies (O’Flaherty & Phillips, 2015). However, the effectiveness of flipped classrooms is constrained by students’ motivation to learn. On one hand, flipped classrooms require students to have high levels of autonomous learning and self-management skills (Shen & Chang, 2023), yet current university students display varying levels of learning ability and self-discipline, with some failing to complete pre-class tasks proactively (O’Flaherty & Phillips, 2015). On the other hand, when the instructional design fails to effectively stimulate students’ intrinsic motivation, flipped classrooms may struggle to achieve the desired learning outcomes.

Learning motivation is a key driving force for achieving deep learning. According to Keller’s ARCS motivation model, teaching that captures students’ attention, enhances content relevance, builds learning confidence, and provides satisfaction can effectively stimulate and sustain students’ learning motivation (Keller, 1987). Research has preliminarily explored the application of this model in education. For instance, Hew and Lo’s study showed that applying the ARCS model in flipped classrooms significantly increases students’ engagement, self-efficacy, and deep understanding of course content (Hew & Lo, 2018; Karabatak & Polat, 2020). Additionally, the ARCS motivation model has been proven to enhance learning motivation in self-directed learning environments (Goksu & Islam Bolat, 2021). Therefore, integrating the ARCS motivation model into flipped classrooms offers a practical solution to the issue of insufficient motivation activation.

Drawing on existing research, the flipped classroom, as an innovative teaching model, demonstrates significant potential in promoting deep learning. However, its effectiveness largely depends on factors such as capturing students’ attention, enhancing content relevance, building learning confidence, and ensuring learning satisfaction. Although previous studies have explored the impact of flipped classrooms on learning motivation, research on the application of Keller’s ARCS motivation model in this context remains limited. Therefore, this study uses the ARCS motivation model as a framework, focusing on whether the flipped classroom model based on this theory can effectively enhance students’ intrinsic motivation, thereby promoting their deep learning. Through this empirical study, the research aims to provide new theoretical insights and practical guidance for flipped classroom design, particularly in stimulating students’ intrinsic motivation and fostering deep learning.

2. Literature Review

2.1. Deep Learning

Marton and Säljö first introduced the concepts of deep learning and surface learning, discovering through their research that different information processing approaches lead to varying learning outcomes (Marton & Säljö, 1976). Deep learning is an active form of learning based on understanding, emphasizing comprehension, critique, connection, construction, transfer, and application of knowledge. In contrast, surface learning primarily focuses on memorization rather than understanding, often driven by exam preparation and lacking deep engagement or reflection on the inherent meaning of knowledge. Biggs further elaborated on deep learning as a higher cognitive or active learning approach, where students demonstrate understanding, processing, transfer, and critique (Biggs, 1979), providing a clearer framework for defining deep learning.

The theoretical foundations of deep learning are primarily rooted in constructivism and self-determination theory. Constructivism emphasizes learning as a process of constructing new and existing knowledge. According to Piaget, learners assimilate and accommodate new knowledge into their existing cognitive structures, gradually building understanding and application of knowledge (Piaget, 1952). Therefore, deep learning encourages learners to deepen their understanding in the knowledge construction process, enabling them to effectively tackle complex challenges. Self-determination theory, from a psychological perspective, highlights the importance of intrinsic motivation and autonomy in enhancing learning outcomes (Ryan & Deci, 2000). Ekwue and colleagues note that learners with high autonomy and intrinsic motivation are more likely to adopt deep learning strategies, focusing on the deeper meanings of knowledge. This autonomous approach aids learners in achieving sustained and profound mastery of knowledge (Ekwue, 2015).

The occurrence of deep learning is influenced by various factors, with motivation levels, teaching environment, and assessment mechanisms being particularly crucial. First, learners’ motivation, especially intrinsic motivation and autonomy, serves as the core driver of deep learning. Ryan and Deci argue that learners with higher intrinsic motivation are more likely to adopt deep learning strategies, demonstrating greater engagement in their learning (Ryan & Deci, 2020). Therefore, the higher the motivation, the more likely learners are to opt for learning approaches focused on deep understanding. Additionally, the teaching environment plays a significant role in supporting deep learning. For example, problem-based learning has been shown to effectively foster deep learning approaches. Dolmans and colleagues found that problem-based learning encourages learners to explore and resolve real-world problems, deepening their understanding of knowledge (Dolmans et al., 2016). Lastly, the impact of assessment mechanisms on deep learning should not be overlooked. Rushton and Alison noted that timely feedback helps students adjust their learning strategies, and formative assessment not only stimulates learners’ motivation but also promotes the occurrence of deep learning (Rushton, 2005).

Deep learning plays a significant role in promoting cognitive development, motivating students, and enhancing learning outcomes. As a learning approach, it focuses on the formation of durable knowledge structures and high-level thinking abilities through understanding, application, and reflection on knowledge. Specifically, deep learning involves multiple dimensions, such as critical thinking, innovative thinking, active participation, and the integration and application of knowledge. In this process, students not only master the content but also gain the flexibility to transfer and apply knowledge. Research shows that deep learning fosters higher-order thinking skills. For instance, Biggs believed that deep learning promotes critical thinking and creativity, enabling students to better master and apply knowledge (Biggs, 1979). Wolters proposed that students with intrinsic motivation invest more energy and perseverance in deep learning, tending to adopt deep learning strategies and further develop critical thinking (Wolters, 1998). MacFarlane et al. argued that deep learning encourages students to focus on the essence of the problem and integrate new knowledge into their existing knowledge systems, facilitating knowledge transfer (MacFarlane et al., 2006). Ng and Lo suggested that deep learning can significantly enhance learners’ cognitive engagement and task involvement, improving learning outcomes in the classroom (Ng & Lo, 2022). Evans and Cuffe’s study showed that innovative teaching designs, such as peer teaching, can significantly improve students’ learning outcomes and promote deep learning (Evans & Cuffe, 2009). Meanwhile, Rushton and Alison found that formative assessments, as a feedback mechanism in teaching, play an active role in motivating students and deepening their understanding of the learning content (Rushton, 2005).

In summary, while motivation plays a central role in deep learning, there is a gap in research regarding the theoretical frameworks that explain this process, especially the applicability of different motivational theories in fostering deep learning.

2.2. Flipped Classroom Teaching (FC)

The flipped classroom is a modern teaching approach that contrasts with traditional instruction. Its core principle involves reversing the processes of knowledge delivery and internalization. In a flipped classroom, students acquire foundational knowledge through self-learning activities before class (e.g., watching videos, reading materials), while class time is dedicated to activities such as interaction, discussion, and problem-solving to deepen knowledge internalization (Mazur, 2009). A key characteristic of this model is student engagement and self-directed learning. Research indicates that the flipped classroom encourages students to take more responsibility for their learning, thereby enhancing their self-directed learning skills (Doung-In, 2017). Lai and Hwang further note that in a flipped classroom, students independently plan their learning pace, which increases their engagement and motivation. Moreover, the active learning process encourages students to ask questions, seek feedback, and develop critical thinking skills in class (Lai & Hwang, 2016). Compared to traditional classrooms, the flipped classroom places greater emphasis on teacher–student interaction and collaborative learning (Bishop & Verleger, 2013). Additionally, flipped classrooms require teachers to shift from being one-way knowledge transmitters to facilitators and guides of learning, helping students develop critical thinking and self-directed learning abilities.

The implementation of the flipped classroom is supported by several learning theories. Firstly, constructivist learning theory posits that knowledge is constructed through the interaction between learners’ prior knowledge and new information (Piaget, 1977). In a flipped classroom, students construct knowledge through pre-class self-learning and further deepen and expand it through in-class interactions. Secondly, social learning theory emphasizes learning through observation and imitation in interaction and collaboration. Group discussions and problem-solving activities in flipped classrooms provide opportunities for cooperative learning and peer support (Piaget, 1977). Additionally, flipped classrooms integrate the concept of the Zone of Proximal Development, helping students achieve a higher level of understanding through classroom discussions and teacher guidance (Zimmerman, 2002).

Research indicates that the flipped classroom model has a positive impact on various aspects of student learning. In a comparison of flipped classrooms with traditional teaching methods, Mason et al. found that the flipped classroom allows instructors to cover more content in limited classroom time while improving students’ performance on open-ended design problems. Additionally, students showed strong adaptability to this novel teaching model, supporting its broader application across different courses (Mason et al., 2013). Avery et al. found that the flipped classroom helps most students become more confident and engaged in their learning, perceiving it as a convenient and comfortable teaching approach. However, they also emphasized that educators need to consider students’ professional development needs when implementing the flipped classroom, designing student-centered activities to ensure alignment with students’ learning requirements (Avery et al., 2018). Research by Fulton at Byron High School demonstrated that the flipped classroom significantly increased students’ interest, motivation, and academic performance (Fulton, 2012). Snyder et al. and Mazur conducted action research in history and sociology courses, finding that the flipped classroom enhances students’ interest in learning, promotes collaborative learning, stimulates curiosity, and improves core competencies (Mazur, 2009; Snyder et al., 2014). Similarly, Strayer, an educational technology expert, found that the flipped classroom fosters students’ collaborative and innovative abilities, particularly excelling in writing and creative thinking skills (Strayer, 2012). Shi et al. highlighted that the flipped classroom effectively enhances cognitive outcomes in higher education, with well-designed teaching strategies being critical for its successful implementation (Shi et al., 2020). Hernández-Nanclares and Pérez-Rodríguez found that a carefully designed flipped classroom can significantly increase students’ sense of course involvement and satisfaction (Hernández-Nanclares & Pérez-Rodríguez, 2016).

In conclusion, while research shows the benefits of flipped classrooms, there is a lack of exploration regarding the mechanisms behind these effects, particularly how they influence deeper learning outcomes.

2.3. ARCS Motivational Model Theory

The ARCS model, developed by Keller (1987), provides a framework to enhance student motivation through four key components: Attention, Relevance, Confidence, and Satisfaction (Keller, 1987). Attention involves capturing students’ interest through engaging and challenging content. Research indicates that without attention, students are unlikely to process information in a way that fosters deeper engagement (R. E. Reynolds & Shirey, 1988). Relevance connects the learning material to students’ interests and future goals, which increases motivation by highlighting the practical value of learning (Assor et al., 2002). Confidence is built by designing tasks of appropriate difficulty that allow students to experience success, thus strengthening their belief in their capabilities (Keller, 1987). Research suggests that when students find learning material relevant, they are more likely to develop confidence, which in turn encourages greater engagement with the material (Maclellan, 2014). Satisfaction, the final component, refers to the sense of achievement students feel after accomplishing learning goals. Keller notes that positive feedback and the fulfillment of achievement sustain motivation for future tasks. The ARCS model was selected for this study because it provides a structured and actionable framework for instructional design, explicitly linking motivation to specific teaching strategies. The ARCS model’s effectiveness has been demonstrated across various disciplines. Reynolds et al. applied it in an information literacy course, combining self-learning and interactive classroom segments (K. M. Reynolds et al., 2017). Bhakti et al. integrated the ARCS model into a flipped classroom for middle school physics, designing classroom activities that boosted students’ confidence and participation (Bhakti et al., 2021). Similarly, Kim used the model in business English teaching, enhancing students’ participation and academic performance (Kim, 2020). Jeon combined the ARCS model with differentiated instruction in English reading, using precisely designed interactive sessions and feedback mechanisms to significantly elevate students’ confidence and motivation levels (Jeon, 2021).

In summary, while the ARCS model was shown to significantly improve motivation and learning outcomes, future research could explore its application in fostering deep learning and higher-order thinking skills. Additionally, more studies are needed to understand how the ARCS model influences student engagement in more complex learning environments, such as those involving critical thinking and problem-solving.

3. Research Hypotheses and Theoretical Model

3.1. ARCS Motivation Model and the Impact of the Flipped Classroom on Deep Learning

The “Attention” dimension of the ARCS model emphasizes stimulating students’ motivation by enhancing their interest and engagement. The dynamic and interactive nature of the flipped classroom attracts students’ attention, thereby sparking their interest in the learning content. This heightened focus allows students to engage in deeper thinking during class, which promotes the occurrence of deep learning (K. Li & Keller, 2018).

The “Relevance” dimension in the ARCS model underscores the connection between the learning material and students’ real-life experiences or interests. The flipped classroom enables students to select learning resources based on their individual interests and needs, enhancing the relevance of the content. When students perceive that the material is related to their personal goals, interests, or practical applications, they are more likely to engage in deep learning, thereby improving their mastery and understanding of the knowledge (Pi et al., 2021).

The “Confidence” dimension in the ARCS model emphasizes the development of students’ belief in their ability to successfully complete tasks. The flipped classroom supports this by allowing students to learn autonomously and providing immediate feedback, which helps build their confidence. As students gain confidence, they become more actively involved in learning activities, which fosters a deeper understanding and application of the content, thereby improving deep learning outcomes (Meir et al., 2024).

The “Satisfaction” dimension in the ARCS model focuses on students’ sense of achievement and fulfillment after learning. The flipped classroom enhances student autonomy and interaction, allowing them to experience greater satisfaction upon completing learning tasks. This satisfaction motivates students to engage in sustained, deep learning, thereby further improving the quality and effectiveness of deep learning (Chen et al., 2017). Therefore, this study posits the following hypotheses:

H1a. 

Attention has a positive effect on deep learning.

H1b. 

Relevance has a positive effect on deep learning.

H1c. 

Confidence has a positive effect on deep learning.

H1d. 

Satisfaction has a positive effect on deep learning.

3.2. ARCS Motivation Model and the Effect of the Flipped Classroom on Intrinsic Motivation

Based on the teaching mode of the flipped classroom and the four dimensions of the ARCS motivation model, the following hypotheses about the impact of the four dimensions on students’ intrinsic motivation are presented. These hypotheses explore how different dimensions in the flipped classroom stimulate students’ intrinsic motivation, which in turn promotes their learning behavior and learning outcomes.

The “Attention” dimension of the ARCS model emphasizes capturing and maintaining students’ attention by stimulating their interest and creating novel and engaging learning experiences. In the flipped classroom, innovative teaching activities, interactive learning, and engaging content can capture students’ attention and ignite their intrinsic motivation (Zainuddin, 2018).

The “Relevance” dimension in the ARCS model highlights the connection between the learning content and students’ personal interests, needs, experiences, and life backgrounds. The flipped classroom allows students to select learning resources based on their interests and needs, enhancing the relevance of the learning material. When students perceive the content as directly relevant to their lives or interests, they are more likely to engage in autonomous learning, which strengthens their intrinsic motivation (Zhu et al., 2022).

The “Confidence” dimension in the ARCS model emphasizes students’ belief in their ability to complete learning tasks and solve problems successfully. The flipped classroom, through opportunities for self-directed learning, immediate feedback, and collaborative exchanges, helps students build confidence gradually. As students gain mastery over new knowledge and successfully complete tasks, their confidence grows, further enhancing their intrinsic motivation and fueling their enthusiasm for continued learning (F. Li et al., 2024).

The “Satisfaction” dimension in the ARCS model focuses on the sense of fulfillment and achievement during the learning process. In the flipped classroom, the provision of opportunities for autonomous learning, positive feedback mechanisms, and rewards after completing tasks can significantly enhance students’ sense of accomplishment and satisfaction. When students experience emotional satisfaction after completing learning tasks, they are more likely to develop intrinsic motivation, which further stimulates their sustained interest in learning (Stansbie et al., 2013). Therefore, this study posits the following hypotheses:

H2a. 

Attention has a positive effect on intrinsic motivation.

H2b. 

Relevance has a positive effect on intrinsic motivation.

H2c. 

Confidence has a positive effect on intrinsic motivation.

H2d. 

Satisfaction has a positive effect on intrinsic motivation.

3.3. The Impact of Intrinsic Motivation and Deep Learning on Students

Self-determination theory suggests that intrinsic motivation is an individual’s motivation to engage in an activity out of interest or intrinsic fulfillment, rather than external rewards or pressures. Intrinsic motivation drives students to actively engage in learning and to engage in deep thinking and reflection, which in turn drives deep learning (Deci & Ryan, 1985). Deep learning is more than a superficial understanding of knowledge; it includes the development of students’ abilities to understand knowledge deeply, think critically, problem-solve, and apply it creatively. Intrinsic motivation fosters students’ active exploration of learning content and facilitates high-level information processing, thereby enabling deep learning (Zhou et al., 2024). Therefore, this study posits the following hypothesis:

H3. 

Intrinsic motivation has a positive effect on deep learning.

3.4. The Mediating Role of Intrinsic Motivation

As a mediating variable, intrinsic motivation primarily influences learning outcomes and behaviors by enhancing students’ motivation, cognitive depth, and learning strategies. In studies examining the impact of teaching models on learning outcomes, intrinsic motivation often serves as a mediator, linking independent variables (e.g., teaching strategies, learning environments) with dependent variables (e.g., deep learning, academic achievement) (Liang et al., 2018).

The “attention” dimension of the flipped classroom enhances students’ focus on learning content, stimulating their intrinsic interest and motivation. Existing research has shown that intrinsic motivation is a key factor in promoting deep learning. When students’ engagement and interest in learning are heightened, they are more likely to engage in deep learning, such as critical thinking, problem-solving, and in-depth information processing (Kruschke, 2003). The “relevance” dimension of the flipped classroom allows students to connect learning content with their personal interests or practical needs, enhancing their motivation. When students perceive learning as relevant and meaningful to their lives, they are more likely to develop intrinsic motivation, leading to deeper thinking and understanding during the learning process (Kember et al., 2008). The “confidence” dimension of the flipped classroom strengthens students’ self-efficacy and confidence, improving their ability to overcome challenges during the learning process. Intrinsic motivation encourages students to invest more in learning activities, thus promoting their deep learning (Pečiuliauskienė, 2020). The “satisfaction” dimension of the flipped classroom enhances students’ learning satisfaction by providing immediate feedback and boosting their sense of achievement. The interaction between intrinsic motivation and learning achievement encourages students to further engage in deep learning activities, strengthening their learning outcomes (Bailey et al., 2021). Therefore, this study posits the following hypotheses:

H4a. 

Intrinsic motivation mediates the relationship between attention and deep learning.

H4b. 

Intrinsic motivation mediates the relationship between relevance and deep learning.

H4c. 

Intrinsic motivation mediates the relationship between confidence and deep learning.

H4d. 

Intrinsic motivation mediates the relationship between satisfaction and deep learning.

3.5. The Influence of the ARCS Teaching Model in the Flipped Classroom on the Relationship Between Different Dimensions

Based on the ARCS model of motivation, the relationship between the four dimensions (“Attention,” “Relevance,” “Confidence,” and “Satisfaction”) can be further elaborated by examining their interactions and mutual influences. The ARCS model emphasizes the multidimensional nature of motivation, with the four dimensions being interrelated and working together to influence students’ learning motivation (Afjar & Syukri, 2020).

In the FC, teachers who design learning activities that capture students’ interest often link them to students’ practical needs and life experiences. Therefore, enhancing students’ attention to the learning content (“Attention”) also strengthens the relevance of the content to their life context (“Relevance”). When students connect the content to their needs, they are more likely to recognize its meaning and value, boosting their confidence in learning. Increased confidence makes students more likely to complete learning tasks and experience a sense of accomplishment, enhancing learning satisfaction. Confidence is a crucial prerequisite for students to achieve success and satisfaction. Therefore, this study posits the following hypotheses:

H5a. 

Attention has a significant positive effect on relevance.

H5b. 

Relevance has a significant positive effect on confidence.

H5c. 

Confidence has a significant positive effect on satisfaction.

This study first explores the direct impact of the flipped classroom based on the ARCS motivation model on deep learning. Flipped classroom teaching based on the ARCS motivation model was taken as the independent variable, deep learning as the dependent variable, and intrinsic motivation as the mediating variable. The model is shown in Figure 1.

4. Methodology

4.1. Questionnaire Design and Participants

The questionnaire design process involves several steps. First, a thorough review of the literature on flipped classroom teaching models, ARCS motivation theory, and deep learning was conducted. Based on this, a scale developed and revised by scholars like Amabile and Chai was selected as the foundation for the study. Next, the selected established scales were matched with the variables in the study, and based on expert feedback, items with unclear meanings or insufficient measurement value were revised to better align with the research needs. The final version of the questionnaire (see

Table 1. Measurement source.

Variable Name	Subject	Source
Attention (A)	1. In the pre-class micro-lesson videos or learning tasks, I find the content particularly attractive, and it stimulates my interest in learning.	Huang et al. (2019); Loorbach et al. (2015)
	2. In the FC, the thought-provoking or discussion-based questions posed by the teacher effectively stimulated my interest in learning.
	3. In the FC, the teaching videos created by the teacher can continuously capture my attention and make the class content more interesting.
Relevance (R)	1. In the FC, when the teacher clearly explains the importance of the knowledge learned for our professional development or academic achievements, I become more motivated to engage in the current learning tasks.
	2. In the FC, I believe the class time is effectively used for interactive discussions with my classmates, which also enhances my motivation and participation.
	3. In the FC, when I have a sufficient understanding of a particular issue, I proactively participate in discussions and help my groupmates solve problems.
Confidence (C)	1. In the FC, I increasingly want to further study the field of my major.
	2. The class time in the FC helps me more effectively apply the knowledge learned outside of class to real-world problems.
	3. The teacher provides targeted guidance and feedback based on the actual classroom situation, helping me more precisely understand my performance in the learning tasks and their causes.
Satisfaction (S)	1. The interactions and practical activities in the FC helped me gain a deeper understanding and mastery of the theoretical knowledge learned outside of class.
	2. When I receive encouragement and praise from the teacher or my classmates during the learning process, I am more motivated to complete subsequent learning tasks.
	3. From the feedback of the teacher and classmates, I can feel the recognition of my efforts.
Intrinsic Motivation (IM)	1. I am willing to take the initiative to learn new knowledge.	Amabile (1993)
	2. I find it fun to complete various course tasks.
	3. I believe that completing tasks and learning new knowledge brings me joy and satisfaction.
	4. I see overcoming challenges encountered during the learning process as a positive challenge.
Deep Learning (DL)	1. When learning new knowledge, I can relate it to relevant knowledge or experiences from my daily life.	Chowdhry and Osowska (2017)
	2. I usually critically compare different opinions, analyze their validity and potential biases, and form my independent judgment and perspectives based on this analysis.
	3. When encountering complex problems, I will break them down into several smaller ones and solve them one by one to more effectively solve the overall problem.
	4. I regularly reflect on my learning process and outcomes, identify areas of improvement, and take measures to improve to continuously enhance my abilities.
	5. I enjoy exploring and trying new methods or strategies to solve problems.

) consists of two sections: personal demographic information and questions related to the research topic. The first section collects basic personal information, such as gender and grade, with five questions. The second section includes 21 questions related to the research topic, covering six constructs. The questionnaire uses a Likert 5-point scale, with response options ranging from “1” (Strongly Disagree) to “5” (Strongly Agree).

The participants in this study were primarily selected from the eastern, central, and western regions of China, as well as Macau and a few other regions. The selection of these regions is based on several considerations. First, significant differences exist between the eastern, central, and western regions of China in terms of economic development, educational resource allocation, and the application of teaching models. The eastern region is economically advanced, with concentrated educational resources, advanced teaching concepts, and facilities. Students in this region tend to be more engaged and responsive to innovative teaching methods. The central region is in the process of economic development and educational reform, with a relatively balanced distribution of educational resources, and students here are more adaptable to new teaching methods. The western region, though economically lagging and with fewer educational resources, faces unique challenges and demands for educational innovation, supported by policy initiatives. Additionally, Macau has a unique cultural background and education system, which leads to educational needs that differ from those of mainland students. Finally, selecting participants from a few other regions—including international locations and areas beyond the primary target region—helps ensure the diversity and comprehensiveness of the sample.

The survey was distributed online to undergraduate students with experience in flipped classroom teaching by academic staff or administrators from well-known universities. The survey was conducted mainly after the course started, ensuring that participants could accurately reflect their learning experiences and motivation within the learning environment. To ensure the validity of the survey data, detailed instructions on completing the questionnaire, along with ethical approval statements, were provided to participants before the formal survey. Participants were fully informed and consented to participate in the study. This series of measures ensures the reliability and validity of the data, facilitating further in-depth analysis and discussion in the subsequent stages of the research.

4.2. Data Collection and Analysis

This study collected 530 questionnaires via web links and QR codes on the Questionnaire Star platform from October to November 2024. After excluding 35 invalid responses, 495 valid questionnaires were retained, yielding a 92.92% response rate. Partial Least Squares Structural Equation Modeling (PLS-SEM) was employed using Smart PLS 4 to analyze path coefficients and validate the DL model among college students. PLS-SEM, ideal for exploratory research with complex models (Hair et al., 2011), was well-suited for this study, as the valid sample size (495) exceeded the minimum requirement of 300 (MacCallum et al., 1996). Its ability to handle formative and reflective constructs ensured robust analysis of latent variables and relationships in the FC model.

In this study, Partial Least Squares Structural Equation Modeling (PLS-SEM) was used to test the explanatory and predictive power of the model with the help of Smart PLS 4 software and analyze the path coefficients of variables to validate the hypothetical model of the mechanism of influence of deep learning among college students. PLS-SEM offers significant advantages in exploratory and predictive research, making it suitable for studies where the theoretical foundation is still developing, or the model is complex (Hair et al., 2011).

Based on the sample size guidelines developed by MacCallum et al., the minimum required sample size for this study was 300 (MacCallum et al., 1996). However, 495 valid questionnaires were collected, exceeding the minimum requirement and ensuring the robustness of the analysis results. PLS-SEM’s ability to process both formative and reflective constructs makes it suitable for analyzing the multiple latent variables and path relationships influencing deep learning in the FC model.

5. Results

5.1. Descriptive Statistics

Table 2. Descriptive statistics.

	Categories	Frequencies	Percentages (%)
Gender	Male	232	46.8
	Female	263	53.1
School Location	Central Region	158	31.91
	Eastern Region	118	23.83
	Western Region	79	15.96
	Macau	93	18.79
	Others	47	9.49
Place of origin	Macao	53	10.7
	Mainland China	412	83.2
	Others	30	6.0
Grade	Freshman	72	14.5
	Sophomore	101	20.4
	Junior	120	24.2
	Senior	143	28.8
	Others	59	11.9
Major classification	Humanities	113	22.8
	Business and Management	110	22.2
	Science and Engineering	103	20.8
	Education	71	14.3
	Arts and Physical Education	54	10.9
	Medical Sciences	27	5.4
	Others	17	3.4

shows the basic demographic information of the participants. The data indicate that the participants were drawn from universities across different regions, including central, eastern, and western China. Overall, the sample is evenly distributed across the five basic information dimensions and better represents the groups involved in the study.

5.2. Measurement Model Checking

As shown in

Table 3. Assessment of reflective measurement models.

Constructs	Indicators	Factor Loadings	Cronbach’s Alpha	Composite Reliability (rho A)	AVE
Attention (A)	A1	0.792	0.738	0.851	0.656
	A2	0.853
	A3	0.783
Relevance (R)	R1	0.861	0.819	0.892	0.734
	R2	0.867
	R3	0.843
Confidence (C)	C1	0.750	0.718	0.841	0.638
	C2	0.827
	C3	0.818
Satisfaction (S)	S1	0.804	0.790	0.877	0.705
	S2	0.839
	S3	0.874
Intrinsic Motivation (IM)	IM1	0.789	0.801	0.870	0.625
	IM2	0.820
	IM3	0.789
	IM4	0.764
Deep Learning (DL)	DL1	0.656	0.722	0.818	0.573
	DL2	0.717
	DL3	0.705
	DL4	0.703
	DL5	0.656

, the factor loadings for each item corresponding to the latent variables are all greater than 0.65 (Purwanto & Sudargini, 2021), demonstrating the reliability of the indicators. The Cronbach’s Alpha values for each latent variable are all greater than 0.7 (Fornell, 1981), and the composite reliability (CR) values exceed 0.7 (Hair et al., 2021), indicating good internal consistency of the variables. PLS-SEM analysis is typically further assessed by examining convergent reliability and discriminant validity. Convergent reliability requires the average variance extracted (AVE) to be greater than 0.5, which measures the extent to which each item reflects the latent variable (Barclay et al., 1995). All variables meet the specified standard.

Indicators of discriminant validity include the Fornell–Larcker Criterion and the Heterotrait–Monotrait Ratio (HTMT), among others. The results in

Table 4. Fornell–Larcker criterion.

	A	C	DL	IM	R	S
A	0.810
C	0.355	0.799
DL	0.458	0.387	0.688
IM	0.397	0.304	0.500	0.791
R	0.383	0.401	0.407	0.309	0.857
S	0.441	0.308	0.502	0.438	0.425	0.840

show that the square root of the AVE should be greater than the correlation coefficients between the construct and other constructs, establishing the discriminant validity of the model. The cross-loadings in

Table 5. Cross-loadings.

	A	C	DL	IM	R	S
A1	0.792	0.259	0.335	0.290	0.309	0.321
A2	0.853	0.332	0.408	0.390	0.334	0.371
A3	0.783	0.265	0.366	0.275	0.285	0.379
C1	0.246	0.750	0.262	0.192	0.298	0.181
C2	0.343	0.827	0.345	0.259	0.331	0.259
C3	0.257	0.818	0.313	0.271	0.331	0.287
DL1	0.266	0.165	0.656	0.240	0.214	0.336
DL2	0.310	0.281	0.717	0.371	0.281	0.336
DL3	0.360	0.240	0.705	0.355	0.302	0.338
DL4	0.340	0.327	0.703	0.358	0.329	0.350
DL5	0.290	0.293	0.656	0.373	0.258	0.367
IM1	0.387	0.235	0.413	0.789	0.243	0.388
IM2	0.332	0.261	0.443	0.820	0.255	0.354
IM3	0.254	0.225	0.347	0.789	0.255	0.322
IM4	0.267	0.240	0.367	0.764	0.224	0.312
R1	0.348	0.386	0.356	0.273	0.861	0.365
R2	0.340	0.322	0.303	0.245	0.867	0.354
R3	0.295	0.319	0.383	0.274	0.843	0.372
S1	0.320	0.240	0.432	0.348	0.340	0.804
S2	0.365	0.215	0.383	0.356	0.312	0.839
S3	0.420	0.312	0.446	0.395	0.410	0.874

demonstrate that the indicators have higher values on their respective constructs than on other constructs. The HTMT values in

Table 6. HTMT.

	A	C	DL	IM	R
C	0.480
DL	0.620	0.523
IM	0.503	0.396	0.643
R	0.490	0.520	0.521	0.381
S	0.574	0.398	0.662	0.545	0.524

indicate that the HTMT ratio between two variables should be less than 0.85 (Henseler et al., 2016), confirming that the measurement model has good validity.

5.3. Structural Modeling

As shown in

Table 7. VIF.

	VIF
A1	1.453
A2	1.552
A3	1.416
C1	1.361
C2	1.455
C3	1.416
DL1	1.314
DL2	1.390
DL3	1.367
DL4	1.303
DL5	1.231
IM1	1.527
IM2	1.697
IM3	1.701
IM4	1.567
R1	1.795
R2	2.003
R3	1.747
S1	1.484
S2	1.806
S3	1.868

, the collinearity test (VIF) for the hypothesized model in this study is below 3, indicating high collinearity among the indicators and the absence of multicollinearity issues (Kock, 2015). As shown in

Table 8. R² Value.

	R2
C	0.359
DL	0.511
IM	0.456
R	0.545
S	0.493

, the coefficient of determination (R²) is used to indicate the explanatory power of exogenous latent variables on endogenous latent variables. Most R² values are close to or above 0.5, suggesting that the model has good explanatory power overall. However, the latent variable C has an R² value greater than 0.3, indicating that its explanatory variance is relatively low. The relationships between latent variable C and its predictors might be weaker compared to other latent variables, suggesting potential theoretical or contextual differences in how C operates within the given research framework. Despite this, the overall model has good overall explanatory power.

Based on the theoretical model constructed in this study (Figure 1), the data were analyzed using Smart PLS 4 software to generate the structural equation model (Figure 2). The graphical output displays the six constructs in the measurement model, their corresponding indicators, and the relationships among the constructs. Path coefficients are represented by numbered arrows, and composite reliability values are shown at the center of each construct.

The results in

Table 9. Results of hypotheses testing.

Hypotheses	Relationship	p Values	Condition
H1a	A→DL	0.002 **	Support
H1b	R→DL	0.027 *	Support
H1c	C→DL	0.021 *	Support
H1d	S→DL	0.000 ***	Support
H2a	A→IM	0.000 ***	Support
H2b	R→IM	0.218 N.S.	No Support
H2c	C→IM	0.016 *	Support
H2d	S→IM	0.000 ***	Support
H3	IM→DL	0.000 ***	Support
H4a	A→IM→DL	0.002 **	Support
H4b	R→IM→DL	0.253 N.S.	No Support
H4c	C→IM→DL	0.042 *	Support
H4d	S→IM→DL	0.000 ***	Support
H5a	A→R	0.000 ***	Support
H5d	R→C	0.000 ***	Support
H5c	C→S	0.000 ***	Support

Note: *, ** and *** represent p values < 0.05, ≤0.01 and ≤0.001, respectively. N.S. means not significant.

indicate that the flipped classroom model based on the ARCS motivation theory significantly influences deep learning in university students, supporting hypotheses H1a, H1b, H1c, and H1d. Attention, confidence, and satisfaction have a significant positive effect on intrinsic motivation, supporting hypotheses H2a, H2c, and H2d. The results of hypothesis H3 show that intrinsic motivation has a significant positive impact on deep learning, indicating that intrinsic motivation plays a key role in promoting deep learning. Attention, confidence, and satisfaction indirectly influence deep learning through intrinsic motivation, supporting hypotheses H4a, H4c, and H4d. Hypotheses H2b and H4b were not supported, indicating that the direct effect of relevance on intrinsic motivation and its indirect effect on deep learning through intrinsic motivation do not hold in this study.

Additionally, hypotheses H5a, H5b, and H5c correspond to the four motivational dimensions of attention, relevance, confidence, and satisfaction in the ARCS motivation model. These hypotheses tested the relationships among these motivational factors, with results showing significant positive associations between them, supporting the validity of the ARCS motivation model in constructing student learning motivation.

6. Discussion and Conclusions

6.1. Discussion

Based on the ARCS motivation model, this study explores the impact of the flipped classroom on college students’ deep learning. The results show that the four dimensions—attention, relevance, confidence, and satisfaction—positively influence deep learning (H1a–H1d), and attention, confidence, and satisfaction also significantly enhance students’ intrinsic motivation (H2a, H2c, H2d). These findings are consistent with prior studies (An & Qu, 2021; Roehl et al., 2013; Shen & Chang, 2023), and further extend the application of the ARCS model by demonstrating its role in fostering deep learning. They suggest that motivational stimulation in flipped classrooms plays a vital role in shaping learning outcomes, and that well-designed instruction can significantly support deep learning strategies.

Furthermore, intrinsic motivation is found to significantly promote deep learning (H3), confirming the core idea of Self-Determination Theory (SDT) that intrinsic motivation enhances learners’ autonomy, critical thinking, and problem-solving ability (Ryan & Deci, 2020; Zhou et al., 2024). Intrinsic motivation arises when learners’ needs for autonomy, competence, and relatedness are satisfied. This aligns with Thomas and Oldfather’s findings that intrinsic motivation contributes to both the depth and creativity of learning (Thomas & Oldfather, 1997). Pintrich similarly notes that students with high intrinsic motivation are more likely to engage in reflective learning and sustained cognitive effort (Pintrich, 2003). As such, fostering intrinsic motivation should be a central goal in promoting deep learning. Teaching strategies such as providing autonomy-supportive environments, setting challenging goals, and giving positive feedback were shown to effectively stimulate intrinsic motivation (Reeve, 2012).

The findings of this study indicate that relevance does not significantly influence intrinsic motivation, nor does it have a significant indirect effect on deep learning through intrinsic motivation (H2b, H4b). While Keller emphasized relevance as a key motivational factor, recent studies suggest its effect may be moderated by context. In structured environments with low learner autonomy, relevance alone may not trigger sustained motivation (Alamri et al., 2020; Lepper & Malone, 2021). According to SDT, relevance addresses relatedness needs but may fall short in supporting autonomy and competence, which are more crucial for intrinsic motivation (Gopalan et al., 2017; Mackenzie et al., 2018). This supports our finding that relevance does not indirectly affect deep learning via intrinsic motivation. Instead, it may influence learning outcomes through direct cognitive engagement, not motivational commitment (Filgona et al., 2020). Therefore, relevance must be integrated with strategies that also support autonomy and competence to have a stronger motivational impact.

Finally, this study highlights a hierarchical relationship among the ARCS components (H5a–H5c). While the original model treated attention, relevance, confidence, and satisfaction as distinct but complementary (Keller, 1987), our findings suggest a progressive sequence: attention initiates engagement, leading to perceived relevance, which enhances confidence, culminating in satisfaction. This sequence is supported by Expectancy-Value Theory (Awidi & Klutsey, 2024) and Self-Efficacy Theory, and emphasizes that motivation in flipped classrooms evolves through interconnected stages. The findings also highlight that the ARCS-based flipped classroom model is not merely a combination of instructional content and activities, but rather a structured, interdependent system that progressively stimulates and sustains motivation. This model not only illuminates the complex mechanisms underlying learning motivation but also provides a solid foundation and practical guidance for optimizing instructional design and improving learning outcomes.

6.2. Theoretical Contribution

This study provides an in-depth exploration of the internal elements of the ARCS motivation model, revealing a hierarchical relationship among its four components: attention (A) influences relevance (R), relevance (R) influences confidence (C), and confidence (C) affects satisfaction (S). This indicates a relational connection among these components, which differs from Keller’s original model that conceptualized the four elements as distinct but complementary motivational strategies (Keller, 1987). Therefore, the findings of this study expand the research perspective of the ARCS motivation model, highlighting the hierarchical relationships between the components and offering a new interpretation of the model’s theoretical framework.

Secondly, this study applies the ARCS motivation model to the flipped classroom teaching model and explores its mechanism in promoting deep learning through the enhancement of intrinsic motivation. The findings show that the ARCS model promotes deep learning by enhancing students’ intrinsic motivation. This finding confirms that the flipped classroom, as an effective teaching model, not only fosters deep learning but also optimizes the process of motivating students’ engagement (Sergis et al., 2018). However, the study also reveals that the influence of individual elements within the ARCS model (such as relevance) on intrinsic motivation has certain limitations.

Specifically, this limitation may be related to the unique nature of the flipped classroom model. In flipped classrooms, the teaching content typically centers around standardized knowledge delivery (Al-Samarraie et al., 2020). As the flipped classroom emphasizes a student-centered learning approach, students usually acquire learning content independently through predefined resources (Charokar & Dulloo, 2022). As a result, students’ perception of relevance to the content may be relatively low. In contrast to traditional classrooms, where teachers can facilitate flexible interaction and individualized guidance to help students understand the relevance of content to real-life experiences, the autonomous learning nature of the flipped classroom often presents content relevance in a more standardized form (Gureckis & Markant, 2012). This limitation in instructional design may result in weaker effects of “relevance” in motivating students’ intrinsic motivation. Nevertheless, this study demonstrates the overall applicability of the ARCS model’s framework within the flipped classroom teaching model.

Finally, the study shows that after considering the hierarchical relationships among the four components, the ARCS model effectively enhances students’ intrinsic motivation and significantly promotes the occurrence of deep learning. This finding expands the scope of the ARCS model’s application and further emphasizes the critical role of intrinsic motivation in promoting deep learning in university students within the flipped classroom teaching model (Kwong et al., 2024). This not only enriches the theoretical content of the ARCS motivation model but also provides a new reference framework for the instructional design of flipped classrooms.

6.3. Practical Implications

First, teachers must shift their mindset and deeply recognize the significance of the flipped classroom model in modern education. Traditional teaching models often focus on the one-way transmission of knowledge from teacher to student, leading to passive learning, which in turn limits students’ initiative and creativity. The flipped classroom, in contrast, encourages students to engage in preparatory learning through pre-class videos and online resources while using in-class time for in-depth discussions, hands-on activities, and collaborative problem-solving (Mazur, 2009). This model not only enhances students’ classroom engagement and motivation but also develops their self-directed learning abilities and critical thinking skills (Pang, 2022). To implement this model effectively, educators should strategically integrate the four components of the ARCS motivation model—Attention, Relevance, Confidence, and Satisfaction—into instructional design:

Enhancing Attention: Teachers can use highly interactive videos, such as micro-lessons or short scenario-based clips, to spark curiosity and capture interest. For example, in a psychology class, inserting an intriguing question like “Why do people develop herd mentality?” before the lesson can trigger curiosity and sustain attention (Deng & Gao, 2024). This anticipatory approach activates cognitive engagement early, laying the foundation for deeper learning during class. Strengthening Relevance: Course content should be closely tied to students’ professional backgrounds and real-world challenges. In management courses, for instance, discussing real enterprise cases helps students see the practical value of theoretical knowledge (Zhao, 2023). When learners perceive direct relevance to their future careers, their intrinsic motivation increases significantly, reinforcing the connection between academic effort and personal goals. Building Confidence: Scaffolding tasks from simple to complex allows students to build skills progressively, increasing their sense of efficacy. In language learning, for example, transitioning from small-group discussions to class-wide presentations enables learners to gain confidence gradually (Derakhshan et al., 2015). This staged approach is crucial for maintaining engagement, especially for students with lower initial confidence levels. Ensuring Satisfaction: Real-time feedback mechanisms—such as online quizzes, instant polls, and digital reward systems—help students monitor their own progress and feel a sense of accomplishment (Alonso et al., 2023). Satisfaction reinforces sustained motivation and supports long-term commitment to learning tasks.

Additionally, project-based and situational learning can further enhance the flipped classroom by simulating authentic, complex tasks. These activities not only provide context-rich environments but also develop higher-order thinking and collaborative problem-solving skills (Yu et al., 2015). For example, in business courses, students can be tasked with analyzing real-world company strategies and proposing improvements, fostering transferable skills highly valued in the job market. Finally, timely and personalized feedback is essential (Klebba & Hamilton, 2007). Teachers should differentiate feedback according to students’ unique needs and learning profiles. One-on-one tutorials or peer mentoring systems can offer additional support for learners struggling with specific concepts, ensuring that all students can engage meaningfully with the flipped classroom model (Alonso et al., 2023). In sum, these practical strategies underscore that the flipped classroom is more than a shift in content delivery—it is a pedagogical framework that, when informed by motivational theory, enables a more responsive, student-centered, and impactful learning experience.

7. Conclusions

This study, based on the ARCS motivation model, empirically examines the impact of the flipped classroom teaching model on college students’ deep learning. The study found that the flipped classroom teaching model, grounded in the ARCS motivation theory, had a significant positive impact on college students’ deep learning. Specifically, factors such as attention, relevance, confidence, and satisfaction in the flipped classroom all had significant positive effects on deep learning, with some of these effects being mediated by intrinsic motivation. Additionally, this study revealed the hierarchical relationships in the flipped classroom’s promotion of deep learning, such as attention influencing relevance, and relevance influencing confidence. The findings of this study provide a new perspective on how flipped classrooms facilitate deep learning and emphasize the importance of stimulating students’ intrinsic motivation during the teaching process. This not only enriches theoretical research in the fields of flipped classrooms and deep learning but also offers scientific guidance for educational practice. Future research could further explore the application of the ARCS motivation model in various teaching environments and investigate how technological tools can optimize the effectiveness of flipped classroom teaching.

8. Research Limitations and Future Recommendations

Building on previous research, this study explores the impact of the flipped classroom teaching model based on the ARCS motivation theory on college students’ deep learning, providing a deeper understanding of the relationship between the two. The findings not only enrich the theoretical exploration of the ARCS motivation model in flipped classrooms but also offer practical insights for promoting college students’ deep learning. However, due to certain limitations in both subjective and objective conditions, this study has some shortcomings and offers suggestions for future research.

First, the limitation of the sample scope restricts the generalizability of the research findings. The sample in this study primarily focuses on university students from the eastern, central, and western regions of China, as well as Macau. While it covers different regions, it does not reflect the application of the flipped classroom teaching model in an international context. Students from different academic fields and educational levels may exhibit differences in motivation and deep learning outcomes in a flipped classroom (Danker, 2015). Moreover, the research sample is limited to university students, and future studies should consider including participants from different educational levels, such as high school or postgraduate students, to examine how motivation and learning behaviors may vary across these groups. Future research could expand the sample sources to include more participants from diverse cultural backgrounds and educational contexts to validate the generalizability and applicability of the findings.

Secondly, certain extraneous variables that were not considered in this study may have influenced the results. Factors such as prior knowledge, individual proficiency levels, and goal orientation could impact students’ intrinsic motivation and deep learning outcomes. For example, students with higher prior knowledge may engage differently in flipped classroom activities compared to those with limited background knowledge. Similarly, goal orientation—whether students are motivated by mastery goals or performance goals—could influence their engagement and learning depth (Alamri et al., 2020). Future research could incorporate these variables to provide a more comprehensive understanding of the relationships examined in this study.

Finally, in practical teaching, attention should be given to integrating theoretical knowledge with real-life contexts, fostering students’ critical thinking and innovative abilities. The flipped classroom teaching model in different courses should integrate their unique teaching characteristics, supported by the dimensions of the ARCS model, to construct a distinctive and systematic flipped classroom teaching framework, continuously providing impetus for its development. In future research, it is hoped that educators will further explore and practice the application of the ARCS motivation model in more practical courses, helping students better adapt to flipped classroom activities and, by aligning with the characteristics of the flipped classroom, better leverage its advantages to enhance higher education teaching.