Research on the Influence of E-Learning Quality on the Intention to Continue E-Learning: Evidence from SEM and fsQCA

Abstract

This study explores how Chinese college students perceive the quality of e-learning and its impact on their adoption of online learning. The study develops an impact model of e-learning quality on continuous learning intention based on the stimulus-organism-response framework, technology acceptance model, and information system success model. The model is validated using a structural equation model (SEM) and fuzzy set qualitative comparative analysis (fsQCA) with data collected from 253 college students at a university in Eastern China. The SEM analysis shows that perceived ease of use and perceived usefulness positively influence e-learning continuance intention, while system quality and personalization have a positive impact on perceived ease of use, and learning community positively impacts perceived usefulness. The fsQCA analysis shows that college students’ willingness to continue e-learning is not solely dependent on a single factor of e-learning service quality but is also influenced by the interaction between various factors. Therefore, e-learning providers should take into account both external stimuli and internal perception factors when designing e-learning services. The findings of this study have practical implications for improving e-learning quality and enhancing the online learning experience. E-learning providers should consider the importance of system quality, personalization, and learning community in improving perceived ease of use and usefulness, which in turn can increase students’ intention to continue e-learning. The study also highlights the importance of considering the interaction between various factors that influence e-learning adoption rather than relying solely on individual factors.

1. Introduction

Over the last two decades, e-learning has become increasingly popular in higher education due to its flexibility, convenience, and cost-effectiveness [1,2,3]. E-learning platforms offer numerous benefits, such as self-paced learning, personalized instruction, and easy access to learning resources [4,5,6,7,8]. However, some students struggle to maintain their engagement and motivation to continue using these platforms, which poses a significant challenge in e-learning adoption and usage [9,10]. The high dropout rate of e-learning in China, for example, is as high as 15–40% [9]. Therefore, it is essential to understand the factors that influence students’ intention to continue using e-learning platforms to enhance their learning experience and improve the quality of higher education.

This study aims to investigate the factors that affect college students’ intention to continue using e-learning platforms. Previous studies have identified various factors that influence e-learning adoption and usage, such as system quality, information quality, service quality, instructor quality, learning community, personalization, perceived ease of use, perceived usefulness, and more e.g., [11,12,13,14,15,16,17,18,19,20,21]. However, the relationship between the configuration path composed of these factors and the intention of college students to continue using e-learning platforms is still unclear.

To address this research gap, this study will examine the impact of these factors on college students’ intention to continue using e-learning platforms. The theoretical framework of the stimulation-organism-response (S-O-R) framework, technology acceptance model, and Delone and Mclean information system success model will guide the research design. By investigating the factors that affect college students’ intention to continue using e-learning platforms, this study aims to provide insights into the design and development of e-learning services to enhance student engagement and motivation in e-learning.

The remainder of this paper is organized as follows: Section 2 reviews the relevant literature. In Section 3, the research model and hypothesis will be presented. Section 4 will describe the data collection process. Section 5 will present the empirical results of the data analysis. Section 6 will be a discussion of the results, followed by the conclusion in Section 7.

2. Literature Review

2.1. E-Learning Quality

Grönroos [22] pointed out that service quality depends on the gap between the service perceived by users and the service expected by users. The research believes that e-learning is a knowledge service relying on information and communications technology. Its quality is defined as the gap between the user’s psychological expectation before accepting e-learning services and the user’s real and actual experience in receiving services.

Previous research on the quality of e-learning has focused on the learner’s subjective perspective to explore the main factors that affect students’ perception of e-learning quality. Some studies suggest that system quality, information quality, and service quality are the key factors that determine the success of e-learning [13,23,24,25,26]. Lee [17] found that students’ perception of online support learning service quality is a significant predictor of online learning acceptance and student satisfaction through surveys of Korean and American students. Chen and Kuo [27] explored adult learners’ perceptions of e-learning service quality in a commercial environment and they classified e-learning service quality into learning interface, learning community, content, and personalization four dimensions. Wu and Lin’s [28] study showed that factors such as “Curriculum development”, “Evaluation”, “Guidance and tracking”, “Instructional design”, and “Teaching materials” can have a significant impact on the quality of e-learning services. Uppal et al. [6] developed an ELQ model to measure college students’ perception of e-learning service quality and found that “Assurance”, “Responsiveness”, “Tangibility”, “Course Website”, and “Learning Content” were positively correlated with students’ perception of online learning service quality. According to a survey of Vietnamese college students, the influencing factors of online learning service quality are mainly composed of e-learning system quality, e-learning instructor and course materials quality, and e-learning administrative and support service quality [20]. Almushasha and Nassuora [29] conducted an empirical study on 189 learners in Jordanian universities and found that the quality of e-learning services focused on five dimensions: interface design, reliability, responsiveness, trust, and personalization.

2.2. S-O-R (Stimulus-Organism-Response) Framework

The S-O-R (stimulus-organism-response) framework, as the fundamental viewpoint of neo-behaviorist psychology, was developed from the classic S-R (stimulus-response) theory. The “stimulus-response” theory was put forward by John Watson. He believed that psychology should be a way to seek and control behavior, and he described individual behavior as a learned response to external stimuli. Mehrabian and Russell [30] proposed the SOR model, which incorporated the concept of the organism (O) between stimulus (S) and response (R), which aimed to better reflect the cognitive and emotional state of behavior before the response. They emphasized the impact of internal mechanisms on human psychology and behavior. According to the S-O-R model, external stimuli will affect the individual’s cognitive and emotional mechanisms and drive them to act accordingly [31].

In recent years, some scholars have introduced the S-O-R theory into e-learning research and confirmed the rationality of analyzing the behavioral intention of online learners based on the S-O-R model. Zhao et al. [32] believed that the technical environment characteristics of the MOOC system, such as interactivity, media richness, and sociability, would impact students’ intention to continue participating in MOOC courses. Bi [33] surveyed senior high school students who purchased online courses at off-campus tutoring agencies and found that the higher the learner’s perception of teacher support, the stronger their social and psychological resilience in school and the more active they were in participating in online collaborations study. Through research, Yang et al. [34] found that perceived closeness, perceived control, and peer referents positively impact the self-efficacy and well-being of students, making learners more actively participate in e-learning. Yang et al.’s [35] research pointed out that the higher the degree of learning engagement of mobile learners, the stronger their intention to continue learning.

2.3. Technology Acceptance Model

The information technology acceptance model (TAM) was proposed by Davis [36] and has been widely used in research on users’ adoption of information technology or information system. Based on the rational behavior theory, the TAM model focuses on explaining the individual’s attitude and intention to use a specific information system and believes that two main factors determine the individual’s intention to use a system: perceived usefulness and perceived ease of use. Perceived usefulness reflects the extent to which individuals think using a specific system will improve job performance and perceived ease of use reflects the degree to which individuals believe it is easy to use a particular system.

Many researchers have studied learners’ e-learning intentions based on the TAM model. Lee et al.’s [15] research suggested that perceived usefulness and playfulness have a stronger relationship with the intention to use e-learning than with perceived ease of use. Based on the information system success model, technology acceptance model, and self-efficacy theory, Li et al. [37] developed a theoretical framework for the behavioral intention of reusable e-learning systems and found that the service quality, course quality, perceived usefulness, perceived ease of use, and self-efficacy of e-learning have a direct impact on users’ behavioral intention to reuse. Chang and Tung [38] found that compatibility, perceived usefulness, perceived ease of use, perceived systems quality, and computer self-efficacy were crucial factors influencing students’ behavioral intentions to use e-learning course websites.

2.4. The Delone and McLean Model of Information Systems Success

In the early stage of information system research, scholars mainly explained the reasons for users’ adoption of information systems from the perspective of technology acceptance. Still, they lacked adequate tools to measure the success of information systems. Delone and McLean [39] created the information system success model (D&M model) based on analyzing and summarizing the literature on information system success in authoritative information system journals from 1981 to 1987. Delone and McLean [40] improved the D&M model based on absorbing and summarizing the relevant research in the past ten years and put forward the second-generation D&M model. Compared with the first-generation D&M model, the updated D&M model introduces the critical factor of service quality and believes that system quality, information quality, and service quality will affect users’ intention to use the system and satisfaction, affecting net benefits.

Lin [13] and Chen [41] introduced the D&M model when studying the success of e-learning systems, which confirmed the feasibility of the D&M model in explaining users’ perception of e-learning quality. Freeze [42] used the D&M model to examine the success of e-learning systems. The study shows that system and information quality significantly positively affect user satisfaction and system usage. Efiloğlu Kurt [43] examined an e-learning system based on student perceptions using the D&M model. He found that system quality significantly impacts both system usage and user satisfaction, while information quality only significantly impacts user satisfaction. Wang et al. [23] and Holsapple and Lee [44] successively adjusted and modified the D&M model based on summarizing the previous research results, which lead to the development of a research model for evaluating e-learning success from the perspective of learners, and pointed out that the quality of e-learning systems, information quality, and service quality are the decisive factors affecting users’ e-learning adoption and satisfaction.

3. Research Model and Hypotheses

3.1. Research Model

Although the TAM model and D&M model have been widely used and verified as the most influential models in the research of information system adoption, they are not designed and developed for the subdivision field of e-learning. Some scholars have built a hybrid model of TAM and D&M to research e-learning user behavior. For example, Mohammadi [45] analyzed the main factors affecting college students’ use of e-learning platforms by establishing the integration model of TAM and D&M. His research showed that e-learning systems and information quality are the key factors determining college students’ intention to continue e-learning. Ngo et al. [46] assessed the impact of e-learning on students’ learning outcomes by integrating the TAM and D&M models. The study found that students perceived learning outcomes were affected by learning assistance, community-building assistance, and perceived motivation.

To sum up, building a hybrid model by integrating TAM and D&M models is an essential direction for the current research on the sustainability of e-learning. This study uses the S-O-R theory as the research framework. It integrates TAM and D&M models, while taking the characteristic factors of e-learning service quality as stimulus variables, in addition to taking perceived usefulness and perceived ease of use as cognitive and emotional responses that affect individuals in order to explore the internal mechanism between e-learning service quality and users’ e-learning sustainability. The research model is shown in Figure 1.

Stimulus variables are various factors that lead to psychological and behavioral changes when learners accept other factors in e-learning services. This study refers to the D&M model and related research on e-learning quality and believes that in the e-learning environment, the external stimulation received by users mainly comes from the related factors of e-learning quality. This paper divides these factors into six dimensions: system quality (SQ), information quality (IQ), service quality (SQ), instructor quality (InQ), learning community (LC), and personalization (PER). Organism variables refer to the cognitive process that learners are stimulated by factors related to e-learning quality, which mainly includes two dimensions: perceived ease of use (PEU) and perceived usefulness (PU).

3.2. Hypotheses

This study hypothesized that online learners can produce higher perceived usefulness and perceived ease of use (O) in e-learning through their perception of e-learning quality (S) and then form their online continuous learning intention (R).

Consequently, this study proposes the following research hypotheses:

Hypothesis 1 (H1).

SQ positively influences the PEU of e-learning.

Hypothesis 2 (H2).

PER positively influences the PEU of e-learning.

Hypothesis 3 (H3).

IQ positively influences the PU of e-learning.

Hypothesis 4 (H4).

SeQ positively influences the PU of e-learning.

Hypothesis 5 (H5).

InQ positively influences the PU of e-learning.

Hypothesis 6 (H6).

LC positively influences the PU of e-learning.

Hypothesis 7 (H7).

PEU positively influences the PU of e-learning.

Hypothesis 8 (H8).

PEU positively influences the CI of e-learning.

Hypothesis 9 (H9).

PU positively influences the CI of e-learning.

4. Data Collection

The questionnaire designed for web-based learning systems was developed with an extensive review of the literature related to the IS Success model, technology acceptance model, and e-learning [12,17,26,27,36,47,48]. The survey questionnaire uses the Likert 5-point scale to assess the degree of uniformity of the respondents to the measurement items (1 means significantly disagree, 5 means very agree). The study distributed electronic questionnaires to 289 undergraduate and graduate students in a university in Eastern China, and 253 valid questionnaires were recovered. The demographic characteristics of the respondents are shown in

Table 1. Profile of survey respondents (n = 253).

Measure	Items	Frequency	Percentage
Gender	Male	53	20.9
	Female	200	79.1
Education	Undergraduate students	186	73.5
	Master students	67	26.5
Hours of e-learning per week	Less than 0.5 h	38	15.0
	0.5–1 h	71	28.1
	1–3 h	90	35.6
	3–5 h	29	11.5
	5 h or more	25	9.9

.

The study tested the reliability and validity of the scale data, and the results are shown in

Table 2. Summary of measurement scales.

Construct	Measure	Factor Loading	Reference Study
System quality AVE = 0.675 CR = 0.861 Alpha = 0.857
SQ1	The user interface of the e-learning platform is well designed.	0.766	Chiu et al., 2007 [12]
SQ2	The e-learning platform can quickly load all the text and graphics.	0.865
SQ3	The e-learning platform functions well all the time.	0.830
Information quality AVE = 0.670 CR = 0.890 Alpha = 0.886
IQ1	The content of the course materials provided by the e-learning platform is timely.	0.776	Chiu et al., 2007 [12]; Chen and Kuo, 2011 [27]
IQ2	The e-learning platform provides content that exactly fits your needs.	0.869
IQ3	The e-learning platform provides sufficient content.	0.816
IQ4	The content of the course materials provided by the e-learning platform is easy to comprehend.	0.810
Service quality AVE = 0.687 CR = 0.898 Alpha = 0.896
SeQ1	The e-learning platform provides prompt responses to my request.	0.875	Yang et al., 2017 [26]
SeQ2	The e-learning platform provides the right solution to my request.	0.829
SeQ3	The service provided in the e-learning platform attends to an individual’s personalized needs.	0.785
SeQ4	The service provided in the e-learning platform is reliable.	0.823
Instructor quality AVE = 0.653 CR = 0.904 Alpha = 0.902
InQ1	Instructors care about student learning.	0.853	La Rotta et al., 2020 [47]; Pham et al., 2019 [48]
InQ2	Instructors manage to develop student interest in the subject.	0.819
InQ3	Instructors transmit their knowledge clearly.	0.694
InQ4	Instructors respond promptly to questions that students may have outside of class.	0.829
InQ5	Instructors provide an environment that encourages interactive participation.	0.837
Learning community AVE = 0.742 CR = 0.920 Alpha = 0.918
LC1	The e-learning platform makes it easy for you to discuss questions with other users.	0.831	Chen and Kuo, 2011 [27]
LC2	The e-learning platform makes it easy for you to access the shared content from the learning community.	0.885
LC3	The e-learning platform makes it easy for you to discuss questions with experts.	0.823
LC4	The e-learning platform makes it easy for you to share what you learn with the learning community.	0.903
Personalization AVE = 0.764 CR = 0.906 Alpha = 0906
PER1	The e-learning platform enables you to choose what you want to learn.	0.889	Chen and Kuo, 2011 [27]
PER2	The e-learning platform enables you to control your learning progress.	0.868
PER3	The e-learning platform records your learning progress and performance.	0.864
Perceived ease of use AVE = 0.717 CR = 0.910 Alpha = 0.904
PEU1	It is easy for me to skillfully use the e-learning platform.	0.892	Davis, 1989 [36]
PEU2	Operating the e-learning platform does not need to consume too much brain power.	0.838
PEU3	Learning to operate the online system would be easy for me.	0.906
PEU4	I can quickly find the necessary learning resources on the e-learning platform.	0.740
Perceived usefulness AVE = 0.749 CR = 0.922 Alpha = 0.922
PU1	Using an e-learning platform would improve my academic performance.	0.826	Lee, 2010 [17]
PU2	Using an e-learning platform enhances my effectiveness in accomplishing academic tasks.	0.863
PU3	Using an e-learning platform has enhanced my interest in accomplishing academic tasks.	0.879
PU4	I find e-learning platforms useful in my study completion.	0.892
Continuance intention AVE = 0.695 CR = 0.919 Alpha = 0.918
CI1	In the future, I have plans to insist on learning an online course.	0.833	Yang et al., 2017 [26]
CI2	In the future, I plan to finish learning multiple online courses.	0.857
CI3	I would like to spend more time on e-learning in the future.	0.850
CI4	I am willing to use e-learning as a way to improve myself.	0.832
CI5	If I have difficulties in learning, I am still willing to turn to e-learning.	0.793

and

Table 3. Discriminant validity analysis.

	SQ	IQ	SeQ	InQ	LC	PER	PEU	PU	CI
SQ	0.821
IQ	0.612	0.819
SeQ	0.53	0.722	0.829
InQ	0.451	0.649	0.703	0.808
LC	0.435	0.623	0.627	0.725	0.861
PER	0.444	0.692	0.541	0.503	0.603	0.874
PEU	0.473	0.61	0.551	0.458	0.521	0.715	0.847
PU	0.417	0.55	0.526	0.538	0.581	0.575	0.61	0.865
CI	0.421	0.519	0.503	0.462	0.532	0.561	0.59	0.788	0.834

. According to the test results, Cronbach’s alpha coefficient for each factor is higher than 0.8, which means the scale has good reliability. The AVE values of each factor are higher than 0.5, and the CR values are higher than 0.7, which means that the scale has good convergent validity. Meanwhile, the AVE square root values of each factor were higher than the correlation coefficients of the factor with other factors, which indicated that the scale had good discriminant validity.

5. Empirical Results

5.1. Structural Model Analysis

The goodness-of-fit analysis of the measurement model using AMOS 24.0 software (as shown in

Table 4. Results of measurement model fit statistics.

Fit Index	χ²/df	RMSEA	RMR	CFI	NFI	NNFI	TLI	IFI
Recommended value	<3	<0.10	<0.05	>0.9	>0.9	>0.9	>0.9	>0.9
Observed Value	2.159	0.068	0.036	0.917	0.857	0.908	0.908	0.918

) revealed that the values of the fit indicators were within the appropriate criteria, indicating that the model’s fit in this study was good. The results showed that the hypotheses were supported, except for H3, H4, and H5, which did not satisfy the significance criteria, as shown in Figure 2. First, system quality (β = 0.188, p < 0.05) and personalization (β = 0.691, p < 0.001) had a significant positive effect on perceived usefulness, which supported H1 and H2; second, the learning community (β = 0.251, p < 0.05) had a significant positive impact on perceived usefulness, which supported H6; third, perceived ease of use (β = 0.370, p < 0.001) had a significant positive effect on perceived usefulness, which supported H7; finally, perceived ease of use (β = 0.127, p < 0.05) and perceived usefulness (β = 0.778, p < 0.001) had a significant positive effect on the intention to continue learning online, and H8 and H9 were supported.

5.2. Fuzzy-Set Qualitative Comparative Analysis

Before carrying out the qualitative comparative analysis of fuzzy sets, it is necessary to calibrate the questionnaire data and convert Likert scores into fuzzy membership scores. For the calibration of scale data, the calibration process is generally performed using the direct calibration method. Some studies used 5 (full membership), 3 (crossover point), and 1 (full non-membership) as anchors to calibrate the Likert 5-point scale to conform to the original meaning of the questionnaire as much as possible [49,50]. However, when self-report questionnaires involve personal behaviors, emotions, and subjective intentions, respondents tend to be more inclined to engage in positive self-reporting due to social desirability, which often leads to the skewed distribution of questionnaire data [51]. Thus, this study uses the 95% quantile of each condition as the threshold of full membership, the 5% quantile as the threshold of full non-membership, and the 50% quantile as the crossover point [52]. Following Fiss [53], this study adds 0.001 to the column condition with a fuzzy membership score of 0.5 to ensure that no cases are deleted when calculating the fuzzy set.

Although conditional configuration analysis is the core of fsQCA, it is necessary to check whether the conditions meet the criteria of the necessary conditions before this. “Necessary condition” means that the condition must exist in the condition configuration that causes the result to occur. The relationship between the result set and the subset of conditions can be evaluated by analyzing the necessity of the condition. The research uses fsQCA software to analyze the necessity of a single condition variable and obtained two critical indicators of the consistency and coverage of each condition, as shown in

Table 5. Results of necessity analysis.

Variable	High CI		Low CI
	Consistency	Coverage	Consistency	Coverage
SQ	0.794	0.772	0.555	0.728
~SQ	0.720	0.546	0.826	0.843
IQ	0.828	0.787	0.549	0.704
~IQ	0.689	0.531	0.835	0.867
SeQ	0.851	0.707	0.601	0.671
~SeQ	0.604	0.529	0.737	0.870
InQ	0.809	0.723	0.553	0.666
~InQ	0.627	0.510	0.770	0.844
LC	0.857	0.737	0.583	0.675
~LC	0.622	0.526	0.772	0.879
PER	0.898	0.744	0.609	0.679
~PER	0.613	0.538	0.770	0.911
PEU	0.845	0.796	0.530	0.672
~PEU	0.652	0.507	0.838	0.879
PU	0.875	0.802	0.516	0.638
~PU	0.605	0.481	0.840	0.900

Note: ~ indicates the absence of the condition.

. It is generally considered that the criterion for determining the necessary condition is that the consistency of the condition reaches more than 0.9 [54]. Thus, we can see that personalization (consistency ≈ 0.9) is a necessary condition leading to a high level of continuous e-learning intention.

When analyzing the conditional configuration paths that lead to a high level of continuous e-learning intention, considering that the sample is greater than 200, the minimum case frequency threshold is set to 3. Moreover, the raw consistency threshold is set to 0.8 and the PRI consistency threshold is set to 0.75, the truth table is constructed, and the standard analysis is performed in order to output the complex, intermediate, and parsimonious solutions. This study follows the previous fsQCA research convention of considering conditions that appear in both parsimonious and intermediate solutions as core conditions and conditions that appear only in intermediate solutions as peripheral conditions. The output of the two condition configurations is shown in

Table 6. Results of sufficiency analysis.

	Solution (High CI)
Configuration	S1	S2
SQ	●
IQ	⬤	⬤
SeQ	●	●
InQ		●
LC	⬤	⬤
PER	⬤	⬤
PEU	⬤	⬤
PU	⬤	⬤
Raw coverage	0.588	0.598
Unique coverage	0.024	0.034
Consistency	0.975	0.969
Overall solution coverage	0.623
Overall solution consistency	0.969

Note: Black circles (●) indicate the presence of the peripheral condition, large black circles (⬤) indicate the presence of the core condition, and blank indicates that the presence or absence of the condition is insignificant.

.

The sufficiency analysis results show that the consistency of each configuration presented in the table is higher than 0.9, the consistency of the overall solution is 0.969, and the coverage of the overall solution is 0.623, which indicates that the research results have strong explanatory power. The common core conditions of solution 1 and solution 2 are information quality, learning community, personalization, perceived ease of use, and perceived usefulness. In S1 (SQ*IQ*SeQ*LC*PER*PEU*PU), the presence of system quality and service quality are peripheral conditions. In S2 (IQ*SeQ*InQ*LC*PER*PEU*PU), the presence of service quality and instructor quality are peripheral conditions. Comparing the two conditional configurations, the system quality in S1 and the instructor quality in S2 have an alternative relationship.

6. Discussion

The results indicate that both external stimuli factors and internal perception factors influence the continuous use intention of e-learning platforms among college students. External stimuli factors include system quality, information quality, service quality, instructor quality, learning community, and personalization. Meanwhile, internal perception factors refer to perceived ease of use and perceived usefulness. This finding represents an expansion of prior studies that have explored the various factors that can impact a student’s willingness to adopt and utilize e-learning platforms e.g., [15,16,17,21,26,55,56]. By understanding these factors, e-learning service providers may create more engaging and effective learning experiences that ultimately improve student learning outcomes.

Furthermore, the study confirms that perceived ease of use and perceived usefulness positively impact the intention to continue using e-learning platforms, and that perceived ease of use also positively influences perceived usefulness [15,21,45,57,58]. Thus, e-learning service providers should strive to create intuitive and user-friendly platforms that offer seamless and engaging learning experiences for students. The results also highlight that system quality and personalization positively impact perceived ease of use. Therefore, e-learning service providers should design platforms that align with college students’ user interface preferences, offer clear navigation functions, ensure stable and smooth operation, and have strong interaction and management features [44,59,60]. Additionally, they should offer personalized learning options and support that enable students to select their own learning content and customize their learning pace and progress [27,29,61].

Moreover, this study found that the learning community dimension positively impacts the perceived usefulness of e-learning. This finding aligns with previous studies that have shown the positive impact of learning communities on students’ motivation and engagement in online learning [2,5]. To improve the learning community dimension, e-learning service providers should enhance the functions of e-learning discussion, sharing, and answering subject questions [46]. They should also expand the communication channels of e-learning communities through social software tools and encourage members to share their e-learning experiences and exchange knowledge [62].

Finally, the fuzzy set qualitative comparative analysis showed that the current generation of college students’ intention to continue e-learning is not solely due to their perception of a single e-learning service quality factor but also due to the interaction between various factors. This finding highlights the importance of considering multiple factors when developing e-learning services. It suggests that e-learning providers should not only focus on external stimuli factors such as system quality, information quality, service quality, instructor quality, learning community, and personalization but also on internal perception factors such as perceived ease of use and perceived usefulness. By considering both external and internal factors, e-learning providers can create a more comprehensive and effective e-learning service that meets the needs and expectations of their users.

In conclusion, this study provides insights into the factors that influence the continuous use intention of e-learning platforms among college students. The findings suggest that e-learning service providers should consider external stimuli factors and internal perception factors when designing and developing e-learning services. Additionally, they should provide personalized learning options and support for students, improve the functions of e-learning discussion, and expand the communication channels of e-learning communities. These findings have significant implications for the development and implementation of e-learning platforms in higher education.

7. Conclusions

This paper adopted the stimulus-organism-response perspective of new behaviorism psychology and integrated the technology acceptance model and DeLone and McLean information systems success model to explore the decisive factors of college students’ willingness to adopt e-learning continuously. This paper emphasized the importance of e-learning quality in continuous learning intention, which not only expanded the application field of the model but also referenced the significance of the adoption of information systems in other fields, such as e-commerce and e-government. Furthermore, the study used hybrid methods to investigate the complex role of various antecedents, enriching the theoretical results of online learning adoption research. The structural equation model is used to analyze the influencing factors and action paths of the e-learning platform’s continuous use intention, while the fsQCA method identifies the antecedent condition configuration that triggers high continuous e-learning platform use intention. This study found that system quality and personalization had a significant positive impact on perceived ease of use, that the learning community had a significant positive effect on perceived usefulness, and that perceived ease of use and perceived usefulness had a positive effect on the intention to continue using the e-learning platform. The results of the fsQCA analysis showed that both S1 (SQ*IQ*SeQ*LC*PER*PEU*PU) and S2 (IQ*SeQ*InQ*LC*PER*PEU*PU) condition configurations can achieve the high continuous use intention of e-learning platform users.

7.1. Implications for Practitioners

The findings of this study have several practical implications for e-learning service providers. First, they should focus on developing platforms that are user-friendly and easy to navigate, with clear and concise content, robust and reliable system operations, and personalized learning options. Second, they should create a strong sense of community among learners by promoting active participation and collaboration, offering opportunities for feedback and communication, and providing forums for sharing knowledge and experiences. Third, e-learning providers should consider both external and internal factors in developing their services, taking into account students’ preferences, needs, and expectations.

7.2. Limitations and Future Work

This study’s limited focus on college students as the research objects may not fully represent the diverse needs and experiences of different user groups, such as adult learners, K-12 students, and professional learners. Additionally, this study’s lack of multi-level coverage of colleges and universities across the country may limit the generalizability of the findings to different educational contexts. Therefore, future research should aim to include a diverse range of user groups, such as K-12 students, adult learners, and professional learners, to ensure that the findings are representative and applicable to various educational contexts. Moreover, educational contexts vary significantly, and their unique features may affect how students perceive and use e-learning platforms. Thus, future studies should aim to include a range of educational institutions, such as community colleges, four-year colleges and universities, and research institutions, to gain a more comprehensive understanding of the factors that impact the continuous use intention of e-learning platforms across different educational contexts. By doing so, researchers can better identify the unique needs and challenges of various educational settings and develop more effective strategies to support continuous e-learning platform use.