Sage Journals: Discover world-class research

Abstract

This study aims to investigate the influence of interaction on undergraduates’ continuance intention to use Massive Open Online Courses based on the Expectation Confirmation Model (ECM) and to examine the moderating effect of gender. 235 undergraduates were randomly sampled and surveyed, providing self-reports on 10 constructs (student-student interactions, student-teacher interactions, student-content interactions, controllability, responsiveness, confirmation, perceived usefulness, satisfaction, and continuance intention). The Partial Least Squares Structural Equation Modelin was employed for data analysis. The findings indicate that student-teacher interactions, student-content interactions, controllability, and responsiveness have a significant positive impact on undergraduates’ continuance intention to use Massive Open Online Courses. However, student-student interactions do not show a significant relationship with continuance intention. Additionally, no significant gender differences were found concerning the influence of perceived usefulness and satisfaction on continuance intention. The study also discusses the theoretical and practical impacts of undergraduates’ use of Massive Open Online Courses.

Plain Language Summary

Analysing how interaction factors influence university students’ continuance intention for MOOCs based on an extended expectancy confirmation model

1. Aims and purpose of the research There are three research questions: (1)In the MOOCs learning, what are the critical INT elements? (2)How do these key INT elements affect CI? (3)Is there a significant gender difference in undergraduates’ CI towards MOOCs? In order to address the above issues, this study proposes the hypothesis model, which consists of a total of 12 research hypotheses. 2.Background of the research With the rapid advancement of the internet and digital technologies, the global educational landscape is undergoing a profound transformation. Massive Open Online Courses (MOOCs) have emerged within this context, offering learners unparalleled advantages. MOOCs provide not only open, flexible, and high-quality educational resources, allowing learners to access top-tier education regardless of their location, but also enhance the online learning experience significantly due to their rich interactive features and personalized learning paths. Furthermore, MOOCs have established a platform for learners, educators, and industry experts worldwide to communicate and collaborate. These characteristics emphasizing personalization and interactivity distinguish MOOCs from traditional online education methods. However, despite the critical role of interactivity in MOOCs, research on how it specifically influences students’ continuance intention (CI) to participate remains a gap. Thus, delving into the relationship between interaction factors and students’ CI in MOOCs holds significant research value. 3.Methods and research design 3.1 Participants This research garnered data via the Wenjuan Star platform (https://www.wjx.cn/) from three distinguished Chinese institutions: Jiangxi Agricultural University, Jiangxi Normal University, Nanchang Hangkong University. Those engaged in this investigation had undertaken MOOCs coursework for at least one academic term. The samples were selected by random sampling. Through a digital inquiry form, 235 responses were gathered.

Keywords

student-student interactions student-teacher interactions student-content interactions expectation confirmation model continuance intention Massive Open Online Courses

Introduction

The rapid development of internet and digital technologies has reshaped global education, with Massive Open Online Courses (MOOCs) emerging as a transformative model. MOOCs provide open, high-quality resources that enhance accessibility to premier education, irrespective of location (Duru et al., 2021; Tan et al., 2022; Y. Zhang et al., 2021). Unlike traditional online education, MOOCs emphasize interactivity and personalization, fostering engagement among learners, educators, and experts (Alyoussef, 2021; de Moura et al., 2021; Gökdemir et al., 2022).

A key challenge in MOOCs lies in sustaining learners’ engagement, termed continuance intention (CI), which reflects their long-term commitment (Alyoussef, 2023). While research has explored factors influencing CI—such as satisfaction (SAT), motivation (MOT), and perceived usefulness (PU) (He et al., 2023; Mailizar et al., 2021; Y. Wang et al., 2020)—there is limited focus on how specific interaction (INT) dimensions affect CI. In MOOCs, interaction comprises student-teacher interactions (STI), student-student interactions (SSI), student-content interactions (SCI), as well as system-related factors like controllability (CONT) and responsiveness (RES). These interaction elements are essential for enhancing learning experiences and sustaining participation (Mubarak et al., 2021; B. Wu, 2021).

This study also explores the role of gender, which has been recognized as a factor that may influence learners’ engagement and CI. However, whether gender moderates the relationships among key constructs such as PU, SAT, and CI remains underexplored. Addressing these gaps can provide a more nuanced understanding of how interaction influences undergraduates’ use of MOOCs.

This study aims to explore the impact of specific interaction elements—STI, SSI, SCI, CONT, and RES—on undergraduates’ CI to use MOOCs. Furthermore, it investigates whether gender moderates the relationships between PU, SAT, and CI. To achieve these objectives, the study addresses the following research questions:

In MOOCs learning, what are the critical interaction elements influencing CI?

How do these specific interaction elements (STI, SSI, SCI, CONT, and RES) affect CI?

Does gender moderate the relationships between PU, SAT, and CI among undergraduates in MOOCs?

This research makes three key contributions. First, it identifies the specific interaction elements that significantly influence CI in the context of MOOCs, providing a detailed understanding of their effects. Second, it examines the moderating role of gender in the relationships between key constructs such as PU, SAT, and CI, addressing an existing gap in the literature. Third, it offers practical recommendations for the design of effective MOOCs platforms, aiming to enhance student SAT and engagement. These findings contribute to both theoretical advancements in online education and practical improvements in MOOCs system development.

Literature Review

Expectation Confirmation Model

The Expectation-Confirmation Theory (ECT), proposed by Oliver (1980), provides a foundational framework for understanding consumer SAT by examining the relationships among performance, expectations, confirmation (CON), and SAT. Building upon this, Bhattacherjee (2001) introduced the Expectation-Confirmation Model (ECM) within the context of information systems, integrating elements from the Theory of Planned Behavior (TPB), the Diffusion of Innovations Theory, and the Technology Acceptance Model (TAM). The ECM comprises four key components: PU, CON, SAT, and CI.

In the ECM framework, PU reflects the perceived value of a technology or service, CON assesses whether expectations align with the experience, SAT measures user SAT based on system performance and interaction quality, and CI indicates the user’s intent to continue usage. These components are widely applied in user behavior research, influencing educational outcomes, sustained learning, and engagement(Alturki & Aldraiweesh, 2023; Rekha et al., 2022).

Recent studies have expanded the ECM to deepen its application in various contexts, integrating constructs such as PU, attitudes, compatibility, and system quality to explore their influence on CI (Ashrafi et al., 2020; Taghizadeh et al., 2022; H. Yang et al., 2023). These extensions reveal how user motivation, system performance, and interaction quality shape CI, particularly in online learning and educational technologies.

Another significant observation is that many studies have explored the integration of the ECM with other models, expanding on the ECM’s foundation to incorporate theories and models like ISSM (H. Yang et al., 2023), UTAUT (Taghizadeh et al., 2022), TTF (T. Wang, Lin, et al., 2021), TPB (Rajeh et al., 2021), and TAM (Ashrafi et al., 2020). This suggests that researchers attempt to provide a more comprehensive explanation for CI by merging different theoretical models. Over the years, the ECM has been widely regarded as a robust and refined model, extensively applied in various domains, including online learning and distance education (Calvo et al., 2020; Duru et al., 2021; Y. Li et al., 2021), mobile application and technology adoption (Alismaiel et al., 2022; Meng & Li, 2023; Togaibayeva et al., 2022), healthcare (Cheng, 2020), and entertainment and gaming (Rohan et al., 2021; L. Zhang et al., 2021). For instance, Taghizadeh et al. (2022) extended ECM by incorporating constructs from the UTAUT model, revealing that hedonic motivation and facilitating conditions significantly enhance CI through their influence on SAT. Similarly, T. Wang, Lin, et al. (2021) combined ECM with the TTF framework, demonstrating that task-technology fit positively impacts PU and CI, emphasizing the importance of aligning system capabilities with user needs. These studies highlight ECM’s robustness while identifying gaps in understanding how its components, particularly CI, operate in specific contexts. For instance, limited research has examined the role of user interactions (e.g., teacher-student, student-content) and system characteristics (e.g., CONT, RES) in shaping CI within educational technologies. To address these gaps, this study investigates the nuanced mechanisms influencing CI, focusing on the interplay between interaction factors and ECM components in MOOCs.

Interaction

INT refers to the mutual activities, information exchange, and dynamic feedback processes between users and computer systems or applications. It drives user perceptions and SAT with specific learning systems or platforms (Wei et al., 2015). Research on INT has progressed from traditional to modern dimensions, integrating both interpersonal and technological interactions to explore their influence on users’ perceptions of the usefulness and their CI to use e-learning systems (Alshammari & Alshammari, 2024; X. Li et al., 2022; Pan et al., 2024; Yin & Lin, 2022). Moore (1989) identified three kinds of INT types: SSI, STI, and SCI, which have been widely applied in traditional educational contexts to examine their effects on users’ acceptance and intention to adopt e-learning systems or new technologies (Karaoglan Yilmaz & Yilmaz, 2020; Kim & Kim, 2021; X. Li et al., 2022; Whitelock-Wainwright et al., 2021; Q. Yang & Lee, 2021; Yin & Lin, 2022). For example, Yin and Lin (2022) demonstrated that various interaction features in mobile banking—such as human–human interaction, human–information interaction, and human–system interaction—positively and significantly influenced users’ PU and perceived ease of use. Similarly, X. Li et al. (2022) confirmed that social interaction in online learning systems has a significant positive effect on the CI of vocational college students to use these systems.

With the emergence of modern learning platforms such as MOOCs, human-computer interaction has become central to online learning. Human-computer interaction emphasizes dimensions of learner-system interactions, including CONT and RES. CONT refers to the user’s ability to control the learning process, such as adjusting preferences or freely navigating content, while RES describes the system’s timely feedback, such as providing instant help or adaptive responses (Alturki & Aldraiweesh, 2023; Chen, 2023; Y. Li et al., 2021; Salta et al., 2022; Shi et al., 2022; Q. Yang & Lee, 2021). These characteristics not only improve user experience but also strengthen CI through seamless interaction design. Q. Yang and Lee (2021) investigated the role of human-computer interaction in MOOCs, demonstrating that user-friendly system design and clear functionalities significantly enhance SAT. Alturki and Aldraiweesh (2023) highlighted that SSI and STI are equally crucial in boosting learner engagement and academic performance. Additionally, Salta et al. (2022) indicated that SSI, STI, and SCI are significant factors affecting student online learning.

In this study, we categorize interaction into two dimensions: human-computer interaction and interpersonal interaction. Interpersonal interaction encompasses three key dimensions—SSI, STI, and SCI—highlighting their roles in fostering knowledge sharing and collaborative learning. Meanwhile, human-computer interaction emphasizes the importance of designing user-friendly systems with features such as CONT and RES to enhance the overall learning experience and meet user needs.

Research Hypotheses

Student-Student Interactions, Student-Teacher Interactions, Student-Content Interactions and Confirmation

Numerous studies have explored the relationship between INT and CON in online learning (Al Mamun & Lawrie, 2023; Chen, 2023; Kim & Kim, 2021) and e-learning (Cheng, 2020; Owusu-Agyeman et al., 2018; Sumi & Kabir, 2021). Specifically, Cheng (2020) introduced a comprehensive research model based on ECM and flow experience, examining the antecedents of the CI of healthcare professionals towards cloud-based e-learning. The findings indicated that INT among learners significantly influences cloud-based e-learning systems’ CON. In addition, Kim and Kim (2021), as well as Alismaiel et al. (2022), argued that STI contribute to the confidence of learners, making them more willing to invest time and effort in learning, which in turn enhances the CON of their expectations of the learning system. A study by Owusu-Agyeman et al. (2018) indicated that in an e-learning environment, the course content positively impacts e-learners’ cognition. Furthermore, Chen (2023) suggested that active INT between students and learning content can foster a deeper understanding and application of textbooks and multimedia resources. In this study, the more active the INT (SSI, STI, and SCI), the higher the CON of the learning system by students. Based on the aforementioned research findings, the following hypotheses are proposed:

H1a: SSI has a significant positive influence on CON.

H2a: STI has a significant positive influence on CON.

H3a: SCI has a significant positive influence on CON.

Student-Student Interactions, Student-Teacher Interactions, Student-Content Interactions and Perceived Usefulness

Previous research has explored the relationships between SSI, STI, SCI, and PU. INT can influence learners’ perceptions of the actual value of educational technology, courses, and learning content by enhancing the learning experience (B. Wu, 2021), garnering feedback from teachers (Bradley-Dorsey et al., 2022; Owusu-Agyeman et al., 2018), and boosting learning motivation (Çebi, 2023; Togaibayeva et al., 2022). B. Wu (2021) focused on SSI, asserting that such INT can spark interest, foster positive learning experiences, strengthen student engagement, and subsequently enhance their PU of the learning activities. Bradley-Dorsey et al. (2022) underscored the significance of STI in online learning environments, highlighting their considerable impact on the classroom atmosphere and the perceived learning experience. Xianhan et al. (2022) found a significant positive correlation between SCI and PU. Through multimedia resources or online platforms, INT amplifies learners’ motivation and autonomy, reinforcing their acknowledgment of the system’s usefulness. In this study, the more active the INT (student-student, student-teacher, student-content), the higher the students’ PU of the learning system. Therefore, based on the aforementioned research findings, the following hypotheses are proposed:

H1b: SSI has a significant positive influence on PU.

H2b: STI has a significant positive influence on PU.

H3b: SCI has a significant positive influence on PU.

Controllability, Responsiveness, and Confirmation

CONT refers to the extent to which users can control a technology, product, or service (Tan et al., 2022). It describes an individual’s subjective perception of the difficulty of a particular action or behavior. Conversely, RES refers to the degree to which a system rapidly responds to user actions and requirements in an academic context (Purwanto et al., 2020). Previous studies have discussed CONT and RES in e-learning (Çebi, 2023; Y, Li & Shang, 2020; Rajeh et al., 2021; Sumi & Kabir, 2021), health behavior (Hagger et al., 2022), and gaming INT (Xue et al., 2023). Firstly, CONT gives users a sense of autonomy. A study by Rajeh et al. (2021) revealed that if learners perceive an e-learning system to have poor CONT, it might result in dissatisfaction with the system, thus reducing their CON of it. Çebi (2023) also noted that when learners have greater control over the learning system, their expected learning outcomes are likely to align more closely with their actual results. This, in turn, enhances their CON of the system. These studies provide robust evidence of the positive impact of CONT on CON. Secondly, RES ensures the system’s prompt reaction. Sumi and Kabir (2021) researched how online learners evaluate e-learning tools and found that the system’s RES positively impacted learners’ perceptions. Further supporting this notion, a study by Y. Li and Shang (2020) indicated that when a learning system rapidly addresses learners’ needs and answers their questions, their expected CON of the system improves. This encourages long-term utilization of the tool. Specifically, the stronger a student’s perception of system CONT and the system’s RES, the higher their expected CON of the system. Based on these findings, the following hypotheses are proposed:

H4a: CONT has a significant positive influence on CON.

H5a: RES has a significant positive influence on CON.

Controllability, Responsiveness, and Perceived Usefulness

Several scholars have conducted extensive research exploring the effects of CONT on PU (Çebi, 2023; Hagger et al., 2022; Rajeh et al., 2021). For instance, Çebi (2023) that CONT is a primary predictor of PU in e-learning systems. Rajeh et al. (2021) investigated the factors influencing the adoption of e-learning by general practitioners and found a significant correlation between CONT and PU.

Additionally, the influence of RES on PU has also been discussed among scholars (Alturki & Aldraiweesh, 2023; Ndzinisa & Dlamini, 2022; Sumi & Kabir, 2021). According to Alturki and Aldraiweesh (2023), if learners believe the system can rapidly respond to meet their needs and address their queries, they are more inclined to view the system as useful. Xue et al. (2023) examined the moderating role of gamification in social INT domains, and the results revealed that RES positively impacts PU. In this study, the stronger the students’ sense of system CONT and RES, the higher their PU of the system. Therefore, based on the aforementioned research findings, the following hypotheses are proposed:

H4b: CONT has a significant positive influence on PU.

H5b: RES has a significant positive influence on PU.

Constructs of the Expectation Confirmation Model

In the ECM, CON, PU, SAT, and CI are the main constructs. Relationships between these constructs have been well-documented in various studies. Firstly, CON has a significant positive influence on PU and SAT(Ashrafi et al., 2020; Cheng, 2020; Meng & Li, 2023; Rohan et al., 2021; T. Wang, Lin, et al., 2021; H. Yang et al., 2023). For instance, H. Yang et al. (2023) found that CON significantly enhances PU among beginners in blended higher education, while Q. Yang and Lee (2021) observed that higher levels of CON increase students’ SAT in sustainable education MOOCs. Similarly, Rohan et al. (2021) demonstrated that CON significantly impacts SAT by integrating motivation and gamification into the ECM. PU also has a strong positive influence on both SAT and CI, as it directly affects users’ SAT with technology, products, or services (Alturki & Aldraiweesh, 2023; Ashrafi et al., 2020; T. Wang, Lin, et al., 2021; X. Wang, Liu, et al., 2021; J. Zhang et al., 2023). For instance, Ashrafi et al. (2020) identified PU as the strongest predictor of students’ CI toward e-learning systems, while J. Zhang et al. (2023) emphasized its importance in enhancing user SAT within online learning systems. Finally, SAT plays a crucial role in promoting CI, as supported by findings from H. Yang et al. (2023) in a blended learning context and Rohan et al. (2021), who confirmed that SAT significantly affects CI in MOOC environments.

In summary, the ECM emphasizes the interactions among CON, PU, SAT, and CI. These relationships explain users’ acceptance and SAT levels with information systems and effectively predict their CI. Therefore, this study proposes the following hypotheses:

H6: CON has a significant positive influence on PU.

H7: CON has a significant positive influence on SAT.

H8: PU has a significant positive influence on SAT.

H9: PU has a significant positive influence on CI.

H10: SAT has a significant positive influence on CI.

The Moderating Role of Gender

Existing studies have identified gender as a moderating factor in the relationship between PU, SAT, and CI in e-learning systems (X. Li et al., 2022; J. Zhang et al., 2023). Research spans domains such as online education (Shihan Izkair & Modi Lakulu, 2023; T. Wang, Lin, et al., 2021; Q. Zhang et al., 2023), behavioral psychology (Al Mamun & Lawrie, 2023; Şahin et al., 2022), and social interaction (Y. Li & Shang, 2020; L. Zhang et al., 2021). For example, in the context of online education, X. Li et al. (2022) found that males prioritize task completion and performance, making them more influenced by PU in forming CI. Similarly, Kaur and Kaur (2022) observed a stronger correlation between PU and behavioral intention among males, suggesting that gender moderates this relationship. In the context of MOOCs. Additionally, gender differences are evident in SAT. Shihan Izkair and Modi Lakulu (2023) reported that females are more affected by social support, which can enhance SAT in online learning environments. However, these differences depend on the context.

Based on the above studies, we propose the following hypotheses:

H11: Gender positively moderates the relationship between PU and CI.

H12: Gender positively moderates the relationship between SAT and CI.

In this research, the model and hypotheses are depicted in Figure 1.

Figure 1.

Research models and hypotheses.

Methods

Participants

This study collected data through the Wenjuan Star platform (https://www.wjx.cn/) from three distinguished Chinese institutions: Jiangxi Agricultural University, Jiangxi Normal University, and Nanchang Hangkong University. These institutions were selected based on their active involvement in MOOCs, which include diverse course content, such as general education, business, engineering, and arts. The MOOCs varied in format, ranging from fully asynchronous to blended models with scheduled live sessions. Each course typically spanned 14 to 16 weeks per semester, with an average of 2 to 3 hr of online instruction per week. This ensured that the sample was representative of a diverse group of students enrolled in MOOCs across different academic backgrounds.

To participate in this study, students were required to have completed at least one semester of MOOCs coursework, accumulating 28 to 48 hr of online learning. Data collection took place from February to August 2023, yielding 256 responses. To ensure data quality, a rigorous screening process was applied. Incomplete questionnaires were excluded first, followed by an analysis of response times. Questionnaires completed in less than 5 min were deemed invalid. After excluding 21 incomplete or excessively fast responses, 235 valid questionnaires were retained for analysis.

Data Collection

The sampling process followed a stratified random sampling method to ensure representativeness of different academic backgrounds and levels of MOOCs participation. The research team collaborated with instructors from Jiangxi Agricultural University, Jiangxi Normal University, and Nanchang Hangkong University to introduce the survey and randomly select participants. Both mandatory and elective MOOCs were included, ensuring diverse participation. Students in mandatory courses participated as part of their curriculum, while those in elective courses voluntarily registered, motivated by reasons such as earning credits, improving skills, or personal interests.

To confirm the sample’s representativeness, a chi-square test for gender distribution was conducted (male = 43.06%, female = 52.11%; p = .517) (Table 1). The results indicated no significant difference between the sample and the overall population, suggesting a strong representativeness of the sample. This approach ensured randomness and representativeness across various academic disciplines and participation levels.

Table 1.

Demographic Characteristics of the Sample.

Demographic characteristic	Category	Number	Percentage
Gender	Male	85	36.2
Gender	Female	150	63.8
Age	Under 18	3	1.3
	18–20 Years	153	65.1
	20–22 Years	63	26.8
	Over 22 Years	16	6.8
Grade level	Freshman	75	31.9
	Sophomore	85	36.2
	Junior	54	23
	Senior	21	8.9
Major	Humanities	129	54.9
	Social Sciences	84	35.7
	Natural Sciences	22	9.4

Measurements

The assessment is bifurcated into two segments. Initially, we gathered participants’ demographic details such as gender, age, educational level, and discipline. Subsequently, we examined nine elements based on the theoretical framework. To elaborate, the latter section encompasses 27 items, as depicted in Table 2. Employing a 7-point Likert scale, responses could vary from 1 (strongly disagree) to 7 (strongly agree). Each question was sourced directly from pertinent academic works.

Table 2.

Construct Measurement and Source.

Constructs	Items	Measurement	Source
SSI	SSI1	MOOCs allowed me to learn from other students.	Cardoso et al. (2011)
	SSI2	Student interaction is an important part of MOOCs learning.
	SSI3	MOOCs gave me plenty of opportunities to interact with other students.
STI	STI1	My teacher encouraged me to voice my opinion in MOOCs.
	STI2	I had the opportunity to use MOOCs to ask questions.
	STI3	I use MOOCs to interact with teachers.
SCI	SCI1	MOOCs have given me a deeper understanding of what I’m studying.	Y. Li et al. (2021)
	SCI2	MOOCs prompt me to preview and review the learning content in time.
	SCI3	MOOCs have sparked my passion for acquiring knowledge.
CONT	CONT1	Using the MOOCs learning system is completely within my control.	Rajeh et al. (2021)
	CONT2	I have the resources, knowledge, and ability to learn using MOOCs.
	CONT3	Most of my expectations for MOOCs have been confirmed.
RES	RES1	MOOCs were able to provide me with timely service.	Sumi and Kabir (2021)
	RES2	MOOCs can improve my academic performance.
	RES3	MOOCs allow me to gain more knowledge.
CON	CON1	The experience of using MOOCs was better than I expected.	Rohan et al. (2021)
	CON2	The service offered by MOOCs is better than I expected.
	CON3	I feel that I have full control over how I use MOOCs to achieve my learning goals.
PU	PU1	MOOCs can improve my study efficiency.
	PU2	MOOCs can improve my academic performance.
	PU3	MOOCs allow me to gain more knowledge.
SAT	SAT1	I am happy with the learning services offered by MOOCs.	Sumi and Kabir (2021)
	SAT2	I would recommend MOOCs to others.
	SAT3	My choice to use MOOCs was a wise decision.
CI	CI1	I plan to continue using MOOCs in the future.	J. Zhang et al. (2023)
	CI2	I will be using MOOCs regularly in the future.
	CI3	I will be using MOOCs a lot in the future.

Note. SSI = student-student interactions; STI = student-teacher interactions; SCI = student-content interactions; CONT = controllability; RES = responsiveness; CON = confirmation; PU = perceived usefulness; SAT = satisfaction; CI = continuance intention.

Furthermore, we provided detailed descriptions of the items and constructs in the measurement model to enhance transparency and reproducibility. For instance, the construct SSI was measured using three items adapted from Cardoso et al. (2011), including statements like “MOOCs allowed me to learn from other students.” Similarly, STI was measured through items such as “My teacher encouraged me to voice my opinion in MOOCs,” while SCI included items like “MOOCs have given me a deeper understanding of what I’m studying,” adapted from Y. Li et al. (2021). Other constructs, such as CONT from Rajeh et al. (2021), PU, SAT from Sumi and Kabir (2021), and CI from J. Zhang et al. (2023), were similarly operationalized using multiple items adapted from prior literature. These items and their sources are detailed in Table 2 to ensure transparency.

Statistical Analysis

The data was scrutinized employing the Smart Partial Least Square 4.0 (Smart PLS) tool. This software was chosen for its ability to efficiently analyze complex models and latent variables, making it well-suited for this study’s analysis of 9 constructs with 235 samples. Smart PLS is particularly advantageous for its capacity to handle both reflective and formative indicators and its suitability for small to medium sample sizes (Hair et al., 2021).

Results

Measurement Model

In line with the principles outlined by Hair et al. (2021), we executed a procedure to examine the model scales’ reliability and validity. At the outset, we gauged the reliability of every aspect using both indicator loadings and composite reliability (CR). According to Hair et al. (2021), the indicator loadings should surpass 0.708 to ensure satisfactory item reliability. Simultaneously, we employed Cronbach’s alpha to determine the internal consistency of the scales, which ought to be above 0.7. As illustrated in Table 3, all measurements, such as indicator loadings, composite reliability, and Cronbach’s alpha, were above .7, signifying reliable scaling.

Table 3.

Reliability and Convergent Validity.

Constructs	Items	Loadings	Cronbach’s α	CR	AVE
CI	CI1	0.924	.916	0.919	0.855
	CI2	0.92
	CI3	0.93
CON	CON1	0.925	.933	0.934	0.882
	CON2	0.945
	CON3	0.947
CONT	CONT1	0.92	.923	0.924	0.866
	CONT2	0.928
	CONT3	0.944
PU	PU1	0.941	.936	0.936	0.886
	PU2	0.946
	PU3	0.936
RES	RES1	0.891	.907	0.907	0.843
	RES2	0.939
	RES3	0.924
SAT	SAT1	0.928	.921	0.922	0.864
	SAT2	0.926
	SAT3	0.935
SCI	SCI1	0.935	.92	0.921	0.863
	SCI2	0.917
	SCI3	0.935
SSI	SSI1	0.884	.891	0.894	0.822
	SSI2	0.929
	SSI3	0.906
STI	STI1	0.912	.911	0.912	0.849
	STI2	0.935
	STI3	0.918

Following this, we evaluated both these scales’ convergent and discriminant validity. We appraised convergent validity via the Average Variance Extracted (AVE) metric, which needs to be greater than 0.5. We employed the classic Fornell-Larcker approach (Fornell & Larcker, 1981) and cross-loadings benchmarks to determine discriminant validity. As dictated by these benchmarks, the external loading of an indicator on its given construct must outstrip any cross-loadings–essentially, its correlations with alternate constructs (Table 4). Moreover, correlations among the aspects should remain below the AVE’s square root, as illustrated in Table 5. Collectively, this data underscores robust discriminant validity for each aspect.

Table 4.

Discriminant Validity (Cross-Loadings Criterion).

Items	CI	CON	CONT	PU	RES	SAT	SCI	SSI	STI
CI1	0.924	0.735	0.766	0.793	0.707	0.869	0.732	0.729	0.75
CI2	0.92	0.646	0.701	0.718	0.622	0.762	0.704	0.614	0.623
CI3	0.93	0.717	0.745	0.78	0.691	0.792	0.749	0.652	0.698
CON1	0.709	0.925	0.81	0.801	0.845	0.751	0.819	0.761	0.77
CON2	0.698	0.945	0.792	0.83	0.823	0.785	0.779	0.733	0.738
CON3	0.728	0.947	0.814	0.855	0.856	0.816	0.81	0.763	0.792
CONT1	0.769	0.798	0.92	0.757	0.78	0.795	0.849	0.762	0.756
CONT2	0.733	0.779	0.928	0.809	0.782	0.791	0.804	0.771	0.724
CONT3	0.73	0.818	0.944	0.834	0.792	0.797	0.782	0.763	0.708
PU1	0.725	0.862	0.79	0.941	0.816	0.793	0.796	0.702	0.74
PU2	0.788	0.819	0.811	0.946	0.784	0.842	0.784	0.716	0.736
PU3	0.819	0.813	0.826	0.936	0.764	0.884	0.806	0.742	0.721
RES1	0.698	0.796	0.818	0.762	0.891	0.77	0.8	0.755	0.714
RES2	0.69	0.832	0.775	0.776	0.939	0.751	0.772	0.721	0.764
RES3	0.623	0.84	0.731	0.766	0.924	0.703	0.731	0.73	0.718
SAT1	0.762	0.798	0.77	0.828	0.778	0.928	0.74	0.742	0.695
SAT2	0.83	0.783	0.814	0.832	0.738	0.926	0.754	0.712	0.691
SAT3	0.849	0.751	0.795	0.831	0.735	0.935	0.739	0.73	0.685
SCI1	0.716	0.822	0.829	0.784	0.775	0.73	0.935	0.784	0.783
SCI2	0.727	0.778	0.8	0.784	0.786	0.749	0.917	0.741	0.745
SCI3	0.753	0.78	0.798	0.786	0.768	0.752	0.935	0.73	0.753
SSI1	0.594	0.686	0.707	0.652	0.672	0.67	0.683	0.884	0.648
SSI2	0.701	0.755	0.769	0.732	0.759	0.765	0.75	0.929	0.749
SSI3	0.663	0.737	0.758	0.695	0.743	0.691	0.764	0.906	0.804
STI1	0.718	0.759	0.75	0.755	0.72	0.725	0.767	0.787	0.912
STI2	0.689	0.748	0.704	0.693	0.757	0.673	0.741	0.744	0.935
STI3	0.663	0.749	0.709	0.701	0.728	0.653	0.754	0.708	0.918

Note. The boldfaced values represent the factor loadings of the items on their respective constructs.

Table 5.

Discriminant Validity (Fornell-Larcker Criterion).

Constructs	CI	CON	CONT	PU	RES	SAT	SCI	SSI	STI
CI	0.925
CON	0.758	0.939
CONT	0.799	0.858	0.931
PU	0.827	0.883	0.86	0.941
RES	0.73	0.896	0.843	0.837	0.918
SAT	0.876	0.836	0.853	0.893	0.807	0.929
SCI	0.788	0.855	0.871	0.845	0.836	0.801	0.929
SSI	0.722	0.801	0.823	0.765	0.801	0.783	0.809	0.906
STI	0.749	0.817	0.783	0.778	0.797	0.742	0.819	0.811	0.922

Note. The bold numbers on the diagonal are the square root of AVE.

Common Method Variance

A dual-method approach was taken to evaluate the Common Method Variance (CMV). The initial assessment was through Harman’s single-factor test, revealing that no factor dominated the variance (Podsakoff et al., 2003). It was discerned that the predominant factor contributed to 28.127% of the variance, a figure much below the recommended 50% benchmark (Podsakoff et al., 2003). As a secondary measure, the marker variable method was implemented. An unrelated marker variable, one not intrinsically linked to the research model, was incorporated to scrutinize any potential common method variance (Lindell & Whitney, 2001). The maximal variance shared with other factors was a mere 0.0189 (1.89%)—a figure considered minimal (Johnson et al., 2011). Hence, from the outcomes of both evaluations, the presence of a significant common method variance seems unlikely.

Structural Model Evaluation

Collinearity

The Variance Inflation Factor (VIF) values were employed within the predictive constructs to examine collinearity. Hair et al. (2021) posited that optimal VIF values remain beneath 5, but preferably less than 3, to confirm negligible collinearity impact on the model’s estimations. As depicted in Table 6, the VIF values ranged from 2.543 to 4.186, aligning with the prescribed criteria.

Table 6.

VIF Values of Predictor Constructs.

Constructs	CI	CON	PU	SAT
CI
CON			3.965	3.536
CONT		2.543	2.809
PU	2.956			3.536
RES		2.546	3.055
SAT	3.956
SCI		3.584	3.762
SSI		4.183	4.186
STI		3.967	4.112

Significance of the structural model relationship

The significance of the structural model relationship was assessed using the bootstrapping algorithm in Smart PLS. According to Hair et al. (2021), t statistics (t > 1.96), p values (p < .05), and confidence interval (excluding zero) were used to test the significance of the relationship. Table 7 indicates the path coefficient, confidence interval, t statistics, and p values. Specifically, the relationship between PU (β = .221, t = 2.335, p = .020), SAT (β = .679, t = 7.567, p = .000), and CI was positive. Likewise, CON (β = .213, t = 2.353, p = .019) and PU (β = .706, t = 8.582, p = .000) was positively related to SAT. The relationship between CONT (β = .195, t = 2.070, p = 2.070), RES (β = .466, t = 5.741, p = .000), STI (β = .144, t = 2.269, p = .023), and CON was positive. Meanwhile, CON (β = .429, t = 4.464, p = .000), CONT (β = .289, t = 3.440, p = .001), SCI (β = .171, t = 2.016, p = .044) was positively related to PU. Finally, the non-significant relationships were as follows: RES with PU (β = .063, t = 0.735, p = .462), SCI with CON (β = .160, t = 1.686, p = .092), SSI with CON (β = .022, t = 0.293, p = .769), SSI with PU (β = −.040, t = 0.616, p = .538), and STI with PU (β = .044, t = 0.577, p = .564). Therefore, the results of this study support H2a, H4a, H5a, H3b, H4b, H6, H7, H8, H9, and H10, but do not support H1a, H3a, H1b, H2b, H5b, H11, and H12. This indicates that in the context of MOOCs learning, STI, SCI, CONT, RES, CON, PU, and SAT are important interactive elements, all of which significantly and positively influence CI either directly or indirectly. However, gender does not play a moderating role in the relationships between PU and CI, or SAT and CI.

Table 7.

Result of the Significance of the Structural Model Relationship.

Relationships	Original sample	2.50%	97.50%	t Statistics	p	Results
CON → PU	0.429	0.248	0.624	4.464	.000	Supported
CON → SAT	0.213	0.04	0.392	2.353	.019	Supported
CONT → CON	0.195	0.018	0.392	2.070	.039	Supported
CONT → PU	0.289	0.111	0.440	3.440	.001	Supported
PU → CI	0.221	0.04	0.407	2.335	.020	Supported
PU → SAT	0.706	0.539	0.861	8.582	.000	Supported
RES → CON	0.466	0.302	0.620	5.741	.000	Supported
RES → PU	0.063	−0.115	0.226	0.735	.462	Not supported
SAT → CI	0.679	0.501	0.846	7.567	.000	Supported
SCI → CON	0.160	−0.034	0.336	1.686	.092	Not supported
SCI → PU	0.171	0.001	0.336	2.016	.044	Supported
SSI → CON	0.022	−0.111	0.176	0.293	.769	Not supported
SSI → PU	−0.040	−0.161	0.097	0.616	.538	Not supported
STI → CON	0.144	0.024	0.274	2.269	.023	Supported
STI → PU	0.044	−0.105	0.195	0.577	.564	Not supported

Explanatory Power and Predictive Relevance

In the presented model, the explanatory power is delineated by the R² values linked with the endogenous constructs, while the predictive relevance is highlighted by Stone-Geisser’s Q² value (Hair et al., 2021). As reflected in Table 8, the model showcases a commendable capacity for explanation through its R² value. Specifically, the R² value for CI reveals an explanation of 77.7% of the variance therein. Similarly, the variance in SAT is roughly 80.8%, explained by its predictors, as evidenced by its R² value. Furthermore, every Q² value, as presented in Table 8, exceeds zero, underscoring the empirical model’s potent predictive accuracy (Hair et al., 2021).

Table 8.

Explanatory Power and Predictive Relevance.

Constructs	R ²	R ² _Adjusted	Q ²
CI	.777	.775	0.654
CON	.856	.853	0.746
PU	.828	.823	0.722
SAT	.808	.807	0.691

Multi-Group Analysis

Measurement Invariance Analysis

This research utilized a multi-group PLS methodology to discern and evaluate gender differences in path coefficients. Prior studies have widely recognized such a technique (Zhou et al., 2014). Before delving into the moderating impacts, we checked for potential consistency issues using the MICOM procedure (Henseler et al., 2016).

Guided by Hair et al. (2021), undertaking a multi-group analysis requires CON of both structural invariance (i.e., uniform parameters and calculation methods) and compositional invariance (i.e., stable indicator proportions). In Smart PLS 4.0, the former is preset. Meanwhile, the latter is gauged using a permutation method as Hair et al. (2021) outlined. The path frameworks and analytical procedures applied to various gender categories in our research were analogous, setting the groundwork for structural consistency (Henseler et al., 2016). Furthermore, structural invariance was confirmed since both group evaluations operated on identical algorithm configurations (Henseler et al., 2016). Compositional consistency is validated when computed correlation metrics surpass the empirical distribution’s 5% quantile (Henseler et al., 2016). Referencing Table 9, the intrinsic correlations between composite evaluations exceeded the empirical distribution’s 5% threshold, bolstering the argument for compositional consistency (Hair et al., 2021). Consequently, measurement consistency across both sets was verified.

Table 9.

MICOM step 2_Compositional Invariance: Across Males Versus Females.

Construct	Correlations among construct scores (H₀ = 1)	5% Quantile of empirical distribution of c_u	p Value	Compositional invariance?
CI	.999	0.999	.065	Yes
CON	1	1	.336	Yes
CONT	1	1	.341	Yes
PU	1	1	.964	Yes
RES	1	0.999	.791	Yes
SAT	1	1	.48	Yes
SCI	1	1	.928	Yes
SSI	1	0.999	.323	Yes
STI	1	0.999	.898

Multi-Group Analysis

Upon confirming measurement invariance, we conducted a comparative analysis across groups. A binary variable was used to distinguish between genders, referencing males and females (Venkatesh & Morris, 2000). The comparative group analysis in this research assessed paths among distinct groups, as suggested by (Zhou et al., 2014). Table 10 showcases the results from the path coefficient analysis segregated by gender. Data suggests that gender variations in the impact of PU (β_male = .389, β_female = .135, p = .221) and SAT (β_male = .556, β_female = .728, p = .392) on CI are not statistically significant. Thus, the hypotheses previously mentioned are not validated.

Table 10.

Comparison of Path Coefficients (Males and Females).

Relationships	Original (male)	Original (female)	Original difference	2.5%	97.5%	p	Hypotheses
PU → CI	0.389	0.135	0.254	−0.401	0.382	.221	Not Supported
SAT → CI	0.556	0.728	−0.172	−0.370	0.376	.392	Not Supported

Discussion

This research investigates the impact of interaction factors on undergraduates’ CI to use MOOCs and explores the moderating effect of gender. Through a comprehensive review of the existing literature, this study posits that factors such as SSI, STI, SCI, CONT, RES, CON, PU, and SAT significantly influence students’ CI towards MOOCs. The proposed research model’s validity was assessed using Smart PLS. Most hypotheses were confirmed, and collectively, all facets explained 77.7% of the total variance in undergraduates’ CI towards MOOCs’ learning systems. Detailed discussions of the research results concerning the initially proposed research questions will be addressed in subsequent sections.

Student-Student Interactions, Student-Teacher Interactions, Student-Content Interactions, Confirmation

In the context of undergraduates using MOOCs, the effects of SSI, STI, and SCI on CON vary significantly. This study found that SSI did not significantly impact students’ CON, which contrasts with Cheng (2020) findings, where SSI positively influenced the CON of healthcare professionals toward cloud-based e-learning. This inconsistency may arise from differences in engagement levels, as some students actively participate in SSI while others remain passive, diluting the positive effects of interaction. Additionally, the asynchronous nature of MOOCs imposes time and spatial constraints, making effective peer interaction challenging, thereby hindering deep learning and communication. Conversely, STI demonstrated a significant positive impact on CON, consistent with the findings of Alismaiel et al. (2022); Kim and Kim (2021). STI facilitates personalized guidance, timely feedback, and motivation for learning, allowing educators to address individual needs and provide feedback that enhances students’ learning outcomes and motivation, thus strengthening their CON. However, SCI did not significantly affect CON in this study, diverging from Çebi (2023) findings, which highlighted SCI’s positive impact in distance education. This discrepancy could stem from the quality and diversity of MOOCs’ resources or technological limitations. Monotonous or low-quality content and insufficient interactive features may hinder students’ engagement, reducing their CON toward MOOCs. These findings emphasize the nuanced roles of different interaction types in shaping students’ CON in online learning environments.

Student-Student Interactions, Student-Teacher Interactions, Student-Content Interactions, Perceived Usefulness

In the context of undergraduates using MOOCs, the impacts of SSI, STI, and SCI on perceived usefulness (PU) vary significantly. This study found that SSI does not significantly influence PU, contrasting with B. Wu (2021) findings that SSI in MOOC discussion forums enhances students’ PU towards learning activities. This discrepancy may stem from variations in the quality and effectiveness of SSI. Some interactions may involve superficial communication lacking substantive content and support, preventing students from gaining sufficient knowledge and information. Additionally, time and resource constraints might limit students’ ability to fully participate in SSI, diminishing their PU of MOOCs. Similarly, STI does not show a notable influence on PU in this study, diverging from the findings of Bradley-Dorsey et al. (2022), who reported a significant impact of STI in online learning environments. One explanation is that MOOC platforms cater to large student populations, limiting teacher resources and making it challenging to provide personalized interactive experiences. Consequently, students may feel that their interactions with instructors do not align with their learning objectives, reducing their PU for MOOCs. Furthermore, individual differences in learning preferences could influence the effectiveness of STI. Some students may prefer independent learning, while others may require more instructor support, leading to varied perceptions of PU. Moreover, the quality and outcomes of STI may vary across courses, as the nature of instructor interaction and feedback can produce differing results, further diminishing PU for MOOCs. In contrast, SCI demonstrates a significant positive influence on PU, consistent with the findings of Xianhan et al. (2022). SCI facilitates personalized learning, deep learning, and interactive experiences. Through diversified SCI, learners can tailor their education to meet their preferences and needs, fostering a deeper understanding and assimilation of knowledge. This enriched interaction enhances students’ appraisal of the usefulness of learning content in MOOCs, ultimately strengthening their PU. These findings highlight the nuanced roles of SSI, STI, and SCI in shaping students’ perceptions of usefulness in online learning environments.

Controllability, Responsiveness, Confirmation

In the context of undergraduates utilizing MOOCs, both CONT and RES exhibit significant positive impacts on CON, reflecting their essential roles in shaping students’ perceptions of MOOCs. CONT was found to positively influence CON, aligning with Çebi (2023) findings that highlighted the role of online learning readiness and motivation in fostering interaction and enhancing CON in remote learning environments. Two potential reasons explain this impact. Firstly, CONT provides students with a sense of autonomy and proactivity, enabling them to tailor their learning experience by choosing content, timing, and methods that align with their pace and requirements. This autonomy fosters confidence in managing their learning journey, thereby reinforcing their support for MOOCs. Secondly, CONT facilitates personalized learning, allowing students to adapt their learning strategies to their unique circumstances, further enhancing their CON. Similarly, RES significantly influences CON, consistent with the findings of Y. Li and Shang (2020), who demonstrated that system responsiveness positively affects learners’ evaluations of online learning tools. RES meets students’ real-time needs during their learning process; when problems or questions arise, timely solutions and support from the system make students feel their needs are addressed effectively. This responsiveness fosters a sense of reliability and satisfaction, bolstering their approval of MOOCs. Moreover, RES contributes to a positive learning experience by providing swift feedback and assistance, enabling a smoother and more enjoyable learning process, which further strengthens students’ CON towards MOOCs. These findings underscore the importance of both CONT and RES in enhancing students’ confidence and approval of MOOCs.

Controllability, Responsiveness, Perceived Usefulness

When undergraduates use MOOCs, CONT significantly positively affects PU, consistent with the findings of Rajeh et al. (2021), who reported a strong correlation between CONT and PU in e-learning. This can be attributed to two key factors. Firstly, CONT enables students to personalize their learning by tailoring it to their interests and objectives, enhancing engagement and perceived utility. Secondly, CONT fosters motivation and active participation, increasing students’ confidence in MOOCs’ usefulness. In contrast, RES does not significantly impact PU in this study, differing from Sumi and Kabir (2021) findings. This discrepancy could stem from variations in RES quality, with responses occasionally lacking substantive information. Furthermore, RES alone cannot compensate for shortcomings in other areas, such as content quality or interactivity. These findings highlight the distinct roles of CONT and RES in influencing students’ perceptions of MOOCs’ usefulness.

Constructs of the Expectation Confirmation Model

In the context of MOOCs, this study confirms that CON significantly positively affects PU and SAT, consistent with findings by Meng and Li (2023); T. Wang, Lin, et al. (2021); H. Yang et al. (2023). When students’ expectations are met or exceeded, they perceive the system as more useful and experience greater SAT. For instance, Q. Yang and Lee (2021) demonstrated that higher CON enhances SAT by fostering positive emotions and psychological fulfillment. Similarly, PU has a significant positive impact on both SAT and CI, as reported by Alturki and Aldraiweesh (2023). Students who perceive MOOCs as effective in meeting their academic needs are more satisfied and have a stronger intention to continue using them. Moreover, SAT plays a crucial role in influencing CI, aligning with findings by Rohan et al. (2021). Satisfied students are more likely to develop a favorable attitude toward MOOCs, reinforcing their commitment to continued use. These results collectively highlight the interconnectedness of CON, PU, SAT, and CI within the ECM framework. By meeting students’ expectations and demonstrating practical utility, MOOCs can enhance SAT and foster long-term engagement. This underscores the importance of optimizing MOOC design to align with learners’ needs and expectations, thereby sustaining their CI.

The Moderating Role of Gender

In the context of undergraduates using MOOCs, gender does not significantly moderate the relationships between PU, SAT, and CI. This contrasts with findings by X. Li et al. (2022) and Shihan Izkair and Modi Lakulu (2023), who reported significant gender differences in PU and SAT influencing CI in other online learning contexts. Several factors may explain these discrepancies. First, MOOCs primarily focus on academic and professional skills, where PU and SAT are determined more by course quality, interactivity, and platform stability than by gender. Second, increasing gender equality in education and the uniform user experience provided by MOOC platforms reduce the likelihood of gender-specific differences in PU and SAT. Finally, intrinsic motivation to acquire knowledge and skills, which drives most students to use MOOCs, is largely independent of gender. These findings suggest that gender plays a minimal role in moderating the relationships between PU, SAT, and CI in MOOCs.

Implications

This study established a comprehensive model to examine the influence of interaction factors on undergraduates’ CI to use MOOCs. Empirical results accounted for 77.7% of the total variance in the students’ CI to use MOOCs. This research further strengthens the theoretical underpinnings and practical feasibility in online learning. Consequently, the findings have significant theoretical and practical implications.

Theoretical Implications

The research results unveiled several significant findings, which hold substantial implications for the future development of the MOOCs. Firstly, we focus on the performance of the constructed model. This model displays strong predictive power between CON, PU, and SAT and has a moderate prediction for CI. The model explains 85.6%, 82.8%, and 80.8% variances in CON, PU, and SAT in MOOCs learning, respectively. Meanwhile, it exhibits 77.7% explanatory power for CI, which is still substantial, though not as pronounced as the previous three variables. This suggests that the model is highly reliable in interpreting these pivotal variables, and, compared to earlier models, it brings some novelty.

Secondly, the innovative nature of the research theory is highlighted. Under the framework of the ECM, various traditional human-human, human-content, and human-computer INT factors have been incorporated (Al Mamun & Lawrie, 2023; Bradley-Dorsey et al., 2022; Çebi, 2023; Owusu-Agyeman et al., 2018; Q. Zhang et al., 2023), constructing a comprehensive and structured model predicting undergraduates’ CI in MOOCs. On the one hand, our study has refined the research category of MOOCs learning INT, offering a more comprehensive theoretical perspective for the MOOCs learning domain. On the other hand, the rationale behind the extended ECM proposed in this paper has been validated through quantitative analysis. This innovation aids in broadening and enriching interactive theory, offering a more detailed and multi-layered analytical approach to better understand the influence of various types of INT on CI.

Thirdly, our research contribution also lies in its comprehensive analysis of interaction factors. This study integrates multiple interactive elements such as SSI, STI, SCI, CONT, and RES. It unveils the causal relationship between INT and MOOCs’ CI and lays the theoretical foundation for researchers to delve deeper into how learners continuously learn and apply knowledge in MOOCs. This aids in filling the research gap in this field (Y. Li et al., 2021; X. Wang, Liu, et al., 2021; Q. Yang & Lee, 2021).

Lastly, we underscore the innovative aspect of examining group differences. This study, gender is introduced as a potential factor to moderate the relationships among PU, SAT, and CI. Notably, our findings reveal no significant gender differences in how PU and SAT impact CI, and we extensively discuss the reasons behind this outcome. This innovative research fills the void of previous MOOCs CI studies that inadequately considered gender differences.

Practical Implications

The findings of this study offer fresh insights and practical implications for MOOCs platform administrators, educators, and learners.

Firstly, for MOOCs platform administrators, this research underscores the positive influence of STI, SCI, CONT, and RES on undergraduates’ CI to use MOOCs. Educational technology companies and MOOCs learning platforms can enhance INT and personalization by refining course design and interactive features, such as online discussions, Q&A sessions, and real-time tutorials, and by providing personalized learning suggestions and progress-tracking tools (X. Wu & Wang, 2018). These measures can assist students in planning their learning objectives more effectively, thus amplifying their CI of using MOOCs.

Secondly, educators can gain a better understanding of students’ needs and expectations and adjust MOOCs course design and teaching methodologies to increase the appeal and quality of their courses. Emphasizing STI and SCI, proactively participating in online discussions, providing instant feedback, and creating interactive course content can enrich the student’s learning experience.

Additionally, educators can enhance the PU of MOOCs courses by elucidating their practical application and value, fortifying students’ sense of identification with MOOCs, and raising SAT and CI. They can also employ personalized teaching approaches to provide targeted guidance and support according to individual student needs and progress, aiding them in comprehending the course content more effectively.

Lastly, in terms of the learners themselves, the empirical results of this study will facilitate them in engaging more effectively with MOOCs, elevating their SAT and CI. On one hand, learners can optimize their learning strategies by selecting MOOCs courses that offer high-quality interactive opportunities and content, thereby improving their learning experience and SAT levels. On the other hand, learners can sustain their interest and motivation by actively participating in INT with educators and content, fostering enthusiasm for using MOOCs (X. Wu & Tian, 2022). This will assist learners in better tackling the challenges of online learning, improving their learning outcomes, strengthening their CI of using MOOCs, and consequently enhancing their educational attainment and career progression.

Limitations and Future Research Directions

While this study offers a framework for an in-depth investigation into the CI to engage with MOOCs, it is not without limitations. Firstly, the research solely relies on a sample from a specific region in China. This may introduce geographical and cultural biases, consequently limiting the generalizability of the findings. Future studies should consider expanding the sample to include students from diverse geographical and cultural backgrounds to validate the universality of the results. Secondly, this study might not have comprehensively examined other critical interaction factors in MOOCs, such as personalized learning support, assessment feedback mechanisms, and community INT. Future research could delve into the interplay of these elements in MOOC learning. Lastly, regarding the CI to use MOOCs, researchers should not only focus on undergraduates. Incorporating perspectives from teachers, MOOC platform designers, and other stakeholders will also be a worthy avenue of exploration. Considering teachers, students, and designers, this multidimensional approach will help provide a comprehensive understanding of how interaction factors influence undergraduates’ CI to use MOOCs, offering a solid foundation for MOOC design and practice.

Conclusion

In this study, we examined the influence of interaction factors on the CI of undergraduates to engage in MOOCs. Using the ECM as a foundation, we incorporated factors of SSI, STI, SCI, CONT, and RES to construct a structured model for undergraduates’ CI. The empirical findings of this study reveal the following conclusions: STI, SCI, CONT, and RES indirectly positively influence undergraduates’ CI towards MOOCs. CON and PU significantly mediate SAT, directly impacting undergraduates’ CI to engage in MOOCs. SSI does not directly or indirectly affect undergraduates’ CI in MOOCs. Moreover, gender differences do not significantly modulate the relationships among PU, SAT, and CI. Consequently, the study of CI regarding MOOCs is influenced by multiple factors. Understanding the interplay of these factors can assist educators and platform developers in better meeting students’ needs. By continually enhancing the quality of service in MOOC learning systems, we can elevate university students’ CI to MOOCs.

Footnotes

Abbreviations

MOOCs: Massive Open Online Courses, CI: continuance intention, SAT: satisfaction, MOT: motivation, PU: perceived usefulness, INT: interaction, STI: student-teacher interactions, SSI: student-student interactions, SCI: student-content interactions, CONT: controllability, RES: responsiveness, ECT: Expectation-Confirmation Theory, CON: confirmation, ECM: Expectation Confirmation Model, TPB: Theory of Planned Behavior, TAM: Technology Acceptance Model, VIF: Variance Inflation Factor.

ORCID iD

Hongzhu Long

Ethical Considerations

The researchers confirms that all research was performed in accordance with relevant guidelines/regulations applicable when human participants are involved (e.g., Declaration of Helsinki or similar). This study was approved by the Ethics Committee of Chongqing City Vocational College, with the approval number: CCVC-2022-03-0009.

Consent to Participate

Informed consent was obtained from all subjects involved in the study.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Chongqing Municipal Education Commission, NO.: CZ223143; Chongqing Municipal Education Commission, NO.: KJQN2021.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author.

References

Al Mamun

M. A.

Lawrie

(2023). Student-content interactions: Exploring behavioural engagement with self-regulated inquiry-based online learning modules. Smart Learning Environments, 10(1), 1–31. https://doi.org/10.1186/s40561-022-00221-x

Alismaiel

O. A.

Cifuentes-Faura

Al-Rahmi

W. M.

(2022). Online learning, mobile learning, and social media technologies: An empirical study on constructivism theory during the COVID-19 pandemic. Sustainability, 14(18), 11134. https://doi.org/10.3390/su141811134

Alshammari

S. H.

Alshammari

M. H.

(2024). Modelling the effects of emotional engagement and peer interaction on the continuous intention to use asynchronous e-learning. PeerJ Computer Science, 10, e1990. https://doi.org/10.7717/peerj-cs.1990

Alturki

Aldraiweesh

(2023). An empirical investigation into students’ actual use of MOOCs in Saudi Arabia higher education. Sustainability, 15(8), 23, Article 6918. https://doi.org/10.3390/su15086918

Alyoussef

I. Y.

(2021). Massive Open Online Course (MOOCs) acceptance: The role of task-technology fit (TTF) for higher education sustainability. Sustainability, 13(13), 7374. https://doi.org/10.3390/su13137374

Alyoussef

I. Y.

(2023). The impact of Massive Open Online Courses (MOOCs) on knowledge management using integrated innovation diffusion theory and the technology acceptance model. Education Sciences, 13(6), 18, Article 531. https://doi.org/10.3390/educsci13060531

Ashrafi

Zareravasan

Rabiee Savoji

Amani

(2020). Exploring factors influencing students’ continuance intention to use the learning management system (LMS): A multi-perspective framework. Interactive Learning Environments, 30(8), 1475–1497. https://doi.org/10.1080/10494820.2020.1734028

Bhattacherjee

(2001). An empirical analysis of the antecedents of electronic commerce service continuance. Decision Support Systems, 32(2), 201–214.

Bradley-Dorsey

Beck

Maranto

Tran

Clark

Liu

(2022). Is cyber like in-person? Relationships between student-student, student-teacher interaction and student achievement in cyber schools. Computers and Education Open, 3, 100101. https://doi.org/10.1016/j.caeo.2022.100101

10.

Calvo

Lyon

Morales

Wade

(2020). Educating at scale for sustainable development and social enterprise growth: The impact of online learning and a Massive Open Online Course (MOOC). Sustainability, 12(8), 3247. https://doi.org/10.3390/su12083247

11.

Cardoso

A. P.

Ferreira

Abrantes

J. L.

Seabra

Costa

(2011). Personal and pedagogical interaction factors as determinants of academic achievement. Procedia—Social and Behavioral Sciences, 29, 1596–1605. https://doi.org/10.1016/j.sbspro.2011.11.402

12.

Çebi

(2023). How e-learning readiness and motivation affect student interactions in distance learning? Education and Information Technologies, 28(3), 2941–2960.

13.

Chen

(2023). Transactional distance and college students’ learning engagement in online learning: The chain mediating role of social presence and autonomous motivation. Psychology Research and Behavior Management, 16, 2085–2101. https://doi.org/10.2147/prbm.S409294

14.

Cheng

Y.-M.

(2020). Investigating medical professionals’ continuance intention of the cloud-based e-learning system: An extension of expectation–confirmation model with flow theory. Journal of Enterprise Information Management, 34(4), 1169–1202. https://doi.org/10.1108/jeim-12-2019-0401

15.

de Moura

V. F.

de Souza

C. A.

Viana

A. B. N.

(2021). The use of Massive Open Online Courses (MOOCs) in blended learning courses and the functional value perceived by students. Computers & Education, 161, 104077. https://doi.org/10.1016/j.compedu.2020.104077

16.

Duru

Sunar

A. S.

White

Diri

(2021). Deep learning for discussion-based cross-domain performance prediction of MOOC learners grouped by language on FutureLearn. Arabian Journal for Science and Engineering, 46(4), 3613–3629. https://doi.org/10.1007/s13369-020-05117-x

17.

Fornell

Larcker

D. F.

(1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of marketing research, 18(1), 39–50.

18.

Gökdemir

Ö.

Özkan Bayrakçı

Aygün

Hoedebecke

(2022). The perspectives of young general practitioners/family physicians on MOOC as part of continuous education: A descriptive semi-qualitative multinational study. Turkish Journal of Biochemistry, 47(2), 201–207. https://doi.org/10.1515/tjb-2019-0201

19.

Hagger

M. S.

Cheung

M. W.

Ajzen

Hamilton

(2022). Perceived behavioral control moderating effects in the theory of planned behavior: A meta-analysis. Health Psychology, 41(2), 155–167. https://doi.org/10.1037/hea0001153

20.

Hair

J. F.

Jr. Hult

G. T. M.

Ringle

C. M.

Sarstedt

(2021). A primer on partial least squares structural equation modeling (PLS-SEM) (3rd ed.). SAGE Publications, Inc.

21.

Jiang

Zhu

(2023). The influence of educational and emotional support on e-learning acceptance: An integration of social support theory and TAM. Education and Information Technologies (Dordr), 28, 11145–11165. https://doi.org/10.1007/s10639-023-11648-1

22.

Henseler

Ringle

C. M.

Sarstedt

(2016). Testing measurement invariance of composites using partial least squares. International Marketing Review, 33(3), 405–431. https://doi.org/10.1108/IMR-09-2014-0304

23.

Johnson

R. E.

Rosen

C. C.

Djurdjevic

(2011). Assessing the impact of common method variance on higher order multidimensional constructs. Journal of Applied Psychology, 96(4), 744–761. https://doi.org/10.1037/a0021504

24.

Karaoglan Yilmaz

F. G.

Yilmaz

(2020). Learning analytics as a metacognitive tool to influence learner transactional distance and motivation in online learning environments. Innovations in Education and Teaching International, 58(5), 575–585. https://doi.org/10.1080/14703297.2020.1794928

25.

Kaur

(2022). Elucidating the role of gender differences via TAM in e-recruitment adoption in India: A multi-group analysis using MICOM. Bottom Line, 35(2/3), 115–136. https://doi.org/10.1108/bl-11-2021-0104

26.

Kim

D.-J.

(2021). Structural relationship of key factors for student satisfaction and achievement in asynchronous online learning. Sustainability, 13(12), 6734. https://doi.org/10.3390/su13126734

27.

Wang

Wei

(2022). Antecedents of continuance intention in online learning systems among vocational college students: The moderating effect of gender. Frontiers in Psychology, 13, 1088270. https://doi.org/10.3389/fpsyg.2022.1088270

28.

Nishimura

Yagami

Park

H.-S.

(2021). An empirical study on online learners’ continuance intentions in China. Sustainability, 13(2), 889. https://doi.org/10.3390/su13020889

29.

Shang

(2020). Service quality, perceived value, and citizens’ continuous-use intention regarding e-government: Empirical evidence from China. Information & Management, 57(3). 103197. https://doi.org/10.1016/j.im.2019.103197

30.

Lindell

M. K.

Whitney

D. J.

(2001). Accounting for common method variance in cross-sectional research designs. Journal of Applied Psychology, 86(1), 114–121. https://doi.org/10.1037/0021-9010.86.1.114

31.

Mailizar

Burg

Maulina

(2021). Examining university students’ behavioural intention to use e-learning during the COVID-19 pandemic: An extended TAM model. Education and Information Technologies (Dordr), 26(6), 7057–7077. https://doi.org/10.1007/s10639-021-10557-5

32.

Meng

(2023). Understanding Chinese teachers’ informal online learning continuance in a mobile learning community: An intrinsic–extrinsic motivation perspective. Journal of Computing in Higher Education, 36, 275–297. https://doi.org/10.1007/s12528-023-09352-7

33.

Moore

M. G.

(1989). Editorial: Three types of interaction. American Journal of Distance Education, 3(2), 1–7. https://doi.org/10.1080/08923648909526659

34.

Mubarak

A. A.

Ahmed

S. A. M.

Cao

(2021). MOOC-ASV: Analytical statistical visual model of learners’ interaction in videos of MOOC courses. Interactive Learning Environments, 31(5), 3055–3070. https://doi.org/10.1080/10494820.2021.1916768

35.

Ndzinisa

Dlamini

(2022). Responsiveness vs. accessibility: Pandemic-driven shift to remote teaching and online learning. Higher Education Research & Development, 41(7), 2262–2277.

36.

Oliver

R. L.

(1980). A cognitive model of the antecedents and consequences of satisfaction decisions. Journal of Marketing Research, 17(4), 460–469.

37.

Owusu-Agyeman

Larbi-Siaw

Wang

(2018). Exploring the factors that enhance student–content interaction in a technology-mediated learning environment. Cogent Education, 5(1), 1–20. https://doi.org/10.1080/2331186x.2018.1456780

38.

Pan

Ishak

N. A.

Qin

(2024). The application of Moore’s online learning interactions model in learning outcomes: The SOR (stimulus-organism-response) paradigm perspective. Heliyon, 10(7), e28505. https://doi.org/10.1016/j.heliyon.2024.e28505

39.

Podsakoff

P. M.

MacKenzie

S. B.

Lee

J.-Y.

Podsakoff

N. P.

(2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88(5), 879–903. https://doi.org/10.1037/0021-9010.88.5.879

40.

Purwanto

Kuswandi

Fatmah

(2020). Interactive applications with artificial intelligence: The role of trust among digital assistant users. Foresight and STI Governance, 14(2), 64–75. https://doi.org/10.17323/2500-2597.2020.2.64.75

41.

Rajeh

M. T.

Abduljabbar

F. H.

Alqahtani

S. M.

Waly

F. J.

Alnaami

Aljurayyan

Alzaman

(2021). Students’ satisfaction and continued intention toward e-learning: A theory-based study. Medical Education Online, 26(1), 1961348. https://doi.org/10.1080/10872981.2021.1961348

42.

Rekha

I. S.

Shetty

Basri

(2022). Students’ continuance intention to use MOOCs: Empirical evidence from India. Education and Information Technologies, 28(4), 4265–4286. https://doi.org/10.1007/s10639-022-11308-w

43.

Rohan

Pal

Funilkul

Chutimaskul

Eamsinvattana

(2021). How gamification leads to continued usage of MOOCs? A theoretical perspective. IEEE Access, 9, 108144–108161. https://doi.org/10.1109/access.2021.3102293

44.

Şahin

Doğan

Okur

M. R.

Şahin

Y. L.

(2022). Emotional outcomes of e-learning adoption during compulsory online education. Education and Information Technologies, 27(6), 7827–7849.

45.

Salta

Paschalidou

Tsetseri

Koulougliotis

(2022). Shift from a traditional to a distance learning environment during the COVID-19 pandemic: University students’ engagement and interactions. Science & Education (Dordr), 31(1), 93–122. https://doi.org/10.1007/s11191-021-00234-x

46.

Shi

Evans

Shan

(2022). Solver engagement in online crowdsourcing communities: The roles of perceived interactivity, relationship quality and psychological ownership. Technological Forecasting and Social Change, 175, 121389.

47.

Shihan Izkair

Modi Lakulu

(2023). The moderating effects of gender on factors affecting the intention to use mobile learning. Journal of Information Technology Education: Research, 22, 199–233. https://doi.org/10.28945/5094

48.

Sumi

R. S.

Kabir

(2021). Satisfaction of e-learners with electronic learning service quality using the SERVQUAL model. Journal of Open Innovation: Technology, Market, and Complexity, 7(4), 227. https://doi.org/10.3390/joitmc7040227

49.

Taghizadeh

S. K.

Rahman

S. A.

Nikbin

Alam

M. M. D.

Alexa

Ling Suan

Taghizadeh

(2022). Factors influencing students’ continuance usage intention with online learning during the pandemic: A cross-country analysis. Behaviour & Information Technology, 41(9), 1998–2017.

50.

Tan

Liu

Tang

Ding

(2022). Concepts and framework of controllable characteristic evaluation and bidirectional interaction technology of flexible load. Frontiers in Energy Research, 10, 1–4. https://doi.org/10.3389/fenrg.2022.917724

51.

Togaibayeva

Ramazanova

Yessengulova

Yergazina

Nurlin

Shokanov

(2022). Effect of mobile learning on students’ satisfaction, perceived usefulness, and academic performance when learning a foreign language. Frontiers in Education, 7, 1–9. https://doi.org/10.3389/feduc.2022.946102

52.

Venkatesh

Morris

M. G.

(2000). Why don’t men ever stop to ask for directions? Gender, social influence, and their role in technology acceptance and usage behavior. MIS Quarterly, 24(1), 115–139. https://doi.org/10.2307/3250981

53.

Wang

Lin

C.-L.

Y.-S.

(2021). Continuance intention of university students and online learning during the COVID-19 pandemic: A modified expectation confirmation model perspective. Sustainability, 13(8), 4586. https://doi.org/10.3390/su13084586

54.

Wang

Liu

Wang

Tian

(2021). Understanding learner continuance intention: A comparison of live video learning, pre-recorded video learning and hybrid video learning in COVID-19 pandemic. International Journal of Human–Computer Interaction, 38(3), 263–281. https://doi.org/10.1080/10447318.2021.1938389

55.

Wang

Dong

Zhang

(2020). Improving MOOC learning performance in China: An analysis of factors from the TAM and TPB. Computer Applications in Engineering Education, 28(6), 1421–1433. https://doi.org/10.1002/cae.22310

56.

Wei

H.-L.

Lin

K.-Y.

H.-P.

Chuang

I.-H.

(2015). Understanding the intentions of users to “stick” to social networking sites: A case study in Taiwan. Behaviour & Information Technology, 34(2), 151–162.

57.

Whitelock-Wainwright

Tsai

Y.-S.

Drachsler

Scheffel

Gašević

(2021). An exploratory latent class analysis of student expectations towards learning analytics services. The Internet and Higher Education, 51, 100818. https://doi.org/10.1016/j.iheduc.2021.100818

58.

(2021). Influence of MOOC learners discussion forum social interactions on online reviews of MOOC. Education and Information Technologies, 26(3), 3483–3496. https://doi.org/10.1007/s10639-020-10412-z

59.

Tian

(2022). Predictors of entrepreneurship intention among students in vocational colleges: A structural equation modeling approach. Frontiers in Psychology, 12, 797790.

60.

Wang

(2018). Selection of cooperative enterprises in vocational education based on ANP. Educational Sciences: Theory & Practice, 18(5), 1–9.

61.

Xianhan

Chun

Mingyao

Caixia

(2022). Associations of different types of informal teacher learning with teachers’ technology integration intention. Computers & Education, 190, 104604. https://doi.org/10.1016/j.compedu.2022.104604

62.

Xue

Niu

Liang

Yin

(2023). Unraveling the effects of voice assistant interactions on digital engagement: The moderating role of adult playfulness. International Journal of Human–Computer Interaction, 40(17), 4934–4955.

63.

Yang

Cai

Yang

H. H.

Wang

(2023). Examining key factors of beginner’s continuance intention in blended learning in higher education. Journal of Computing in Higher Education, 35(1), 126–143. https://doi.org/10.1007/s12528-022-09322-5

64.

Yang

Lee

Y.-C.

(2021). The critical factors of student performance in MOOCs for sustainable education: A case of Chinese universities. Sustainability, 13(14), 1–22. https://doi.org/10.3390/su13148089

65.

Yin

L.-X.

Lin

H.-C.

(2022). Predictors of customers’ continuance intention of mobile banking from the perspective of the interactivity theory. Economic Research-Ekonomska Istraživanja, 35(1), 6820–6849. https://doi.org/10.1080/1331677x.2022.2053782

66.

Zhang

Liu

Zhang

(2023). What are the key drivers to promote continuance intention of undergraduates in online learning? A multi-perspective framework. Frontiers in Psychology, 14, 1121614. https://doi.org/10.3389/fpsyg.2023.1121614

67.

Zhang

Shao

Feng

(2021). Gamification and online impulse buying: The moderating effect of gender and age. International Journal of Information Management, 61, 102267. https://doi.org/10.1016/j.ijinfomgt.2020.102267

68.

Zhang

Solis

Mukuni

(2023). An exploratory study of learner characteristics, perception of interaction, and satisfaction in online consumer finance courses. Journal of e-Learning and Knowledge Society, 19(1), 36–42.

69.

Zhang

Z. S.

Zhang

Shi

(2021). The relationships of parental responsiveness, teaching responsiveness, and creativity: The mediating role of creative self-efficacy. Frontiers in Psychology, 12, 748321. https://doi.org/10.3389/fpsyg.2021.748321

70.

Zhou

Jin

X.-L.

Fang

(2014). Moderating role of gender in the relationships between perceived benefits and satisfaction in social virtual world continuance. Decision Support Systems, 65, 69–79. https://doi.org/10.1016/j.dss.2014.05.004

Understanding the Impact of Interaction Elements on MOOCs Continuance Intention Among Undergraduates: An Extended Expectation Confirmation Model

Abstract

Plain Language Summary

Keywords

Introduction

Literature Review

Expectation Confirmation Model

Interaction

Research Hypotheses

Student-Student Interactions, Student-Teacher Interactions, Student-Content Interactions and Confirmation

Student-Student Interactions, Student-Teacher Interactions, Student-Content Interactions and Perceived Usefulness

Controllability, Responsiveness, and Confirmation

Controllability, Responsiveness, and Perceived Usefulness

Constructs of the Expectation Confirmation Model

The Moderating Role of Gender

Methods

Participants

Data Collection

Measurements

Statistical Analysis

Results

Measurement Model

Common Method Variance

Structural Model Evaluation

Collinearity

Significance of the structural model relationship

Explanatory Power and Predictive Relevance

Multi-Group Analysis

Measurement Invariance Analysis

Multi-Group Analysis

Discussion

Student-Student Interactions, Student-Teacher Interactions, Student-Content Interactions, Confirmation

Student-Student Interactions, Student-Teacher Interactions, Student-Content Interactions, Perceived Usefulness

Controllability, Responsiveness, Confirmation

Controllability, Responsiveness, Perceived Usefulness

Constructs of the Expectation Confirmation Model

The Moderating Role of Gender

Implications

Theoretical Implications

Practical Implications

Limitations and Future Research Directions

Conclusion

Footnotes

Abbreviations

ORCID iD

Ethical Considerations

Consent to Participate

Funding

Declaration of Conflicting Interests

Data Availability Statement

References