4.1. Number of Publications, Countries, and Institutions
Figure 1 is obtained by statistics from the collected literature, which shows the number of published studies on influencing factors of online learning behavior in the past decade. From 2013 to 2017, the number of studies developed steadily. After 2017, the number of documents on influencing factors of online learning behavior continued to increase, and the growth rate has accelerated in the past two years, especially in 2021, where the number of studies on influencing factors of online learning behavior increased dramatically.
The number of studies on the influencing factors of online learning behavior has increased dramatically. One of the reasons for the enormous growth is that Internet technology has driven the development of online education. Online learners have a variety of learning channels. The teaching methods of various schools have also changed from traditional classroom teaching to mixed teaching, that is, the combination of online and offline. One of the crucial reasons for the sharp increase in research in 2021 is the impact of COVID-19, which forces teaching to go online. That is also consistent with the statements of Al-Kumaim and Navarro et al. [
18,
19].
By analyzing the countries that have studied the influencing factors of online learning behavior,
Figure 2 was obtained. Among them, China ranked first with 181 published articles, accounting for 27.2% of the total published articles in ten years, the United States ranked second with 76 pieces, accounting for 11.4%, and Saudi Arabia and Malaysia ranked third and fourth with 51 and 42 articles, respectively.
Table 1 lists the top ten countries regarding the number of publications and their centrality.
By analyzing research institutions’ publications on influencing factors of online learning behavior,
Figure 3 was obtained. National Taiwan Normal University ranked first with 18 published articles, King Saud University ranked second with 14 published articles, and Beijing Normal University ranked third with 10 published articles.
Table 2 lists the top five institutions with the highest number of publications and their centrality.
However, from the perspective of the centrality of research countries and research institutions, these countries and institutions are not closely linked. This indicates that their cooperation is not close, possibly due to different educational methods and concepts in different countries.
4.2. Keyword Co-Occurrence and Cluster Analysis
Keywords can quickly and accurately reflect the theme and focus of a paper, and their frequency in the literature of a specific research field can reflect the topics that researchers pay attention to together in a certain period, that is, the hot topics in the whole research field. Research on keywords helps analyze the burning issues in a specific scientific area. First, the documents in the WoS folder are imported into CiteSpace to remove duplicate samples. Then, due to a large number of pieces, pruning is selected to cut the network to improve the readability of the network. Finally,
Figure 4 shows the co-occurrence of keywords in the WoS literature. A node represents a keyword, and the larger the node’s size, the more frequently the keyword appears, indicating that the keyword is a research hotspot in this field to some extent. When two keywords appear at the same time, there is a line. It can be seen from
Figure 4 that the high-frequency keywords of WoS research on influencing factors of online learning behavior include online learning, user acceptance, satisfaction, higher education, model, technology acceptance model, education, student, etc. “Centrality” is an indicator to measure the importance of a node in the network. A high centrality value indicates that the more keywords associated with it, the greater the link effect and the higher the importance. Therefore, by comparing the centrality and frequency of keywords, we can find out more accurately and deeply the research hotspots of the influencing factors of online learning behavior and sort out the top 10 keywords in terms of frequency and the top 10 keywords in terms of centrality and centrality, as shown in
Table 3 and
Table 4.
According to the frequency and centrality of the keywords, the research hotspots of online learning behavior influencing factors in WoS literature include online learning, user acceptance, satisfaction, technology acceptance model, higher education, etc. It can be seen from
Figure 4 that some keywords are closely related and some are sparsely related. To explore the profound relationship between these keywords, you can use the clustering function of CiteSpace. The clustering results are shown in
Figure 5. The modular degree Q in the clustering results is 0.7377, much higher than 0.3. Generally, a clustering modular value (Q value) of >0.3 means that the clustering structure is significant—the contour value S = 0.8851 is higher than 0.7. A clustering average contour value (S value) of >0.5 means that clustering is reasonable, and S > 0.7 that clustering is convincing. As such, the structure of this clustering is remarkable and effective.
A total of 22 clusters have been formed in the clustering process, but not every cluster is effective. These 22 clusters need to be screened. The screening criteria are the number of cluster members and the S value. The number of cluster members should be greater than 10 at first, and then the S value should be greater than 0.7. An S value > 0.7 indicates that the internal compactness of the cluster members is good, the inner members are similar, and the clustering is convincing. Finally, 15 clusters are left after screening. These are clustered according to title, and the labels are as follows: #0 e-learning system, #1 student course selection, #2 learning ineffectiveness, #3 planned behavior, #4 user engagement, #5 extended technology acceptance model, #6 different factor, #7 user experience, #8 e-learning service quality influence, #9 systematic review, #10 influence assessment, #11 academic emotion, #12 cross-sectional study, #13 digital transformation, and #14 learning motivating factor.
Through the analysis of cluster members, the clustering of online learning behavior influencing factors can be further divided into two groups. The first group is different online learning behaviors, including #0, #1, #2, #3, #4, #7, and #11. Online learning behavior includes positive online learning behavior and negative online learning behavior. Positive online learning behavior contains two types: online learning engagement and continuous learning [
20]. The engagement behavior of online learning users can be divided into adopting or accepting online learning [
21] and the degree of online learning investment [
22]. Negative online learning behaviors include online learning procrastination [
23] and truancy [
24].
The second group is the influencing factors, including #5, #6, #8, #9, #10, #12, #13, and #14. Online learning behaviors have various influencing factors, including the influencing factors of online learning satisfaction.
Table 5 summarizes the influencing factors of online learning behaviors.
Online learning engagement behavior refers to the extent to which learners are willing to adopt or accept online learning and invest time in online learning. In addition to the influence of COVID-19, the use of information technology by schools is on the rise. Mobile devices and the Internet are widely spread among students. Therefore, online learning will become an essential tool for students. The most basic is the willingness to accept and adopt online learning and the degree of investment in online learning [
38,
39]. Continuous learning behavior refers to the willingness of online learners to continue to use and use online learning systems. Schools will adopt online learning, and learners will learn through some online learning systems, such as MOOCs. The most important thing for these online learning systems is to enable users to continue using them [
31]. Procrastination in online learning refers to online learners’ procrastination, slackness, and anxiety when they are not under face-to-face supervision. Under the influence of COVID-19 closed management, the psychological state of learners will also be greatly affected, resulting in online learning delays [
40]. Online learning truancy is more serious. Some learners will not participate in the course directly, and some will carry out activities unrelated to learning without opening the camera, such as leaving their seats, playing games, browsing other web pages, etc.
In addition to some common factors, such as perceived usefulness, perceived ease of use, quality, and other factors, some scholars have also done corresponding research on some unique factors—for example, differences in individual characteristics, gender, personality, age, etc. In addition, some factors need to be paid attention to. For instance, Luiz Antonio Joia verified that teachers’ digital ability on the technical platform and metacognitive support in the digital environment are important factors for online courses to achieve their teaching goals successfully [
41]. Tsai discussed the influence of metacognition on online learning interest and persistence [
42]. Metacognition refers to implicit or explicit information generated by individuals on their cognition and coping strategies [
43]. The results show that improving learners’ metacognition helps enhance online learning interest and continuous learning through MOOC.
4.3. Empirical Analysis: Key Literature Analysis
The co-citation analysis helps us understand the critical literature in this research field. When another piece at the same time cites two articles, there is a co-citation relationship between the two pieces. The key literature analysis further explains how different online learning behaviors are affected by factors.
Figure 6 shows the cited network of the literature on the influencing factors of online learning behavior. Literature by Al-Fraihat D (2020) has the highest cited frequency, totaling 31 times.
Table 6 lists the top five pieces of literature cited most frequently and their centrality.
Among the five highly cited papers, the first analyzes the critical success factors of accepting online learning and online learning satisfaction through the structural equation model. The factors involved include technical system quality, information quality, service quality, support system quality, learner quality, teacher quality, and perceived usefulness. Perceived usefulness is the critical determinant of acceptance of online learning. If students think the e-learning system is helpful, they are more likely to use it. Another factor, technical system quality, helps overall satisfaction and perception of system effectiveness. However, there is no significant impact on the use of online learning systems because students still use the specific e-learning platform adopted by the school, regardless of its quality. Secondly, the quality of the education system, such as the existence and interactivity of communication tools, the diversity of learning styles, and the provision of evaluation materials (such as tests and assignments) to students, have a significant impact on the use of e-learning systems. The aspects related to teacher quality are closely related to students’ perceived satisfaction and usefulness of the system but have nothing to do with their use. The second literature adopts the structural equation model and takes collaboration quality, information quality, and user-perceived satisfaction as the influencing factors of e-learning use behavior. And the driving factors of user-perceived satisfaction are information quality, system quality, teachers’ attitude towards e-learning, diversity of evaluation, and learners’ perception of interaction with others.
The third piece of literature is an interview on subject analysis through NVivo software, which mainly analyzes the key influencing factors of using online learning during the popularity of COVID-19. The key factors that need to be considered are technical, e-learning system quality, trust, self-efficacy, and cultural factors. Among the quality factors of the e-learning system, perceived ease of use has a significant relationship with the use of online learning. First, if students find the plan challenging, they may lose confidence in it. Secondly, it meets the needs of students and can also be effectively adopted and used. Trust factors include system protection, information privacy, and system reliability. One of the vital trust factors for students to increase their use of e-learning systems is to provide effective and transparent e-learning activities through e-learning system projects and ensure security and freedom from threats. Self-efficacy is one of the core factors in deciding whether to adopt an e-learning system. To increase the adoption of e-learning systems, it is necessary to ensure that students have high self-efficacy to meet the expected functions. Otherwise, if students have low self-efficacy, it is not easy to achieve learning activities through e-learning systems. Cultural factors are also essential factors. One of the factors that should be implemented to increase the use of e-learning systems is to improve the ICT literacy and skills of e-learning users. Social interaction can also increase their participation.
The fourth and fifth articles both adopt the technology acceptance model. The former analyzes the factors of adopting e-learning, mainly involving self-efficacy, enjoyment, experience, computer anxiety, and subjective norms. The latter is to analyze the influencing factors of the continuous use of online learning. The main factors involved include perceived usefulness and attitude, which are crucial to the constant intention of using MOOCs. In addition, perceived ease of use, task technology suitability, reputation, social cognition, and social impact play an important role in predicting persistent intention.