[7] Zhang et al 39-2 Australasian Journal of Educational Technology, 2023, 39(2). 115 A meta-analysis of the moderating role of prior learning experience and mandatory participation on factors influencing MOOC learners’ continuance intention Min Zhang, Sihong Li Wuhan University, China Yan Zhang University of the Chinese Academy of Sciences, China Retaining learners has been an important issue for massive open online course (MOOC) platforms. Given the different, and even contradictory, conclusions in studies on the continuance intention of MOOC learners, this study selected 53 highly correlated empirical studies published from 2008 to 2022 and constructed a research model based on visual knowledge map analysis. Meta-analysis was applied to identify the key factors, and subgroup analysis was conducted to explore the moderating effect of mandatory participation and prior learning experience. The results show that attitude and satisfaction play the most significant role. Perceived usefulness, perceived ease of use, confirmation, social influence, perceived enjoyment, outcome expectation, self-efficacy and task- technology fit all play essential functions, while the direct impact of social presence requires further research. Prior learning experience and mandatory participation have moderating effects on perceived usefulness. MOOC developers should make more efforts and improvements in content quality, social quality and service quality. Implications for practice or policy: • Learners’ continuance intention can be enhanced by improving individual perceived positive feelings related to MOOCs and individual satisfaction with MOOC platforms. • Directors of mandatory courses in MOOCs should place greater emphasis on improving learners’ perceived ease of use of MOOC platforms. • Superintendents of MOOC platforms need to be aware of the role of perceived usefulness of learners with less prior learning experience in their continuance intention. Keywords: MOOC learners, continuance intention, prior learning experience, mandatory participation, meta-analysis Introduction Massive open online courses (MOOCs) refer to learners’ network learning based on MOOCs’ online learning information systems (ISs; Kizilcec et al., 2013). The outbreak of the novel coronavirus disease (COVID-19) has increased the demand for online education (Zayapragassarazan, 2020), indicating new opportunities for educational reform. In such a context, MOOCs, as the core of the international strategic action of educational digitisation, have attracted increasing attention because they can provide an efficient and effective alternative solution for the long-term sustainable development of traditional classroom education during the pandemic. However, for some people, MOOC learning may be a trend. The latest statistics show that the dropout rate of MOOCs is as high as 90% (Gu et al., 2021), which is why scholars have begun to pay increasing attention to learners’ continuous use behaviour. Alemayehu and Chen (2021) recommended that more attention should be paid to investigating learners’ intentions. Therefore, it is of great theoretical value and practical significance to study the continuance intention (CI) of learners with MOOCs. After a systematic literature review of existing studies, three research gaps were identified as follows: Australasian Journal of Educational Technology, 2023, 39(2). 116 First, despite the accumulation of various theories and models on the continuation of technological innovation, such as the expectation confirmation model (ECM) and the technology acceptance model (TAM), the literature abounds in diverse but often contradictory findings. Research includes contradictions on whether the perceived usefulness (PU) of and satisfaction (SAT) with MOOCs are significantly related (Alraimi et al., 2015; Lu et al., 2019). There is also controversy about whether self- efficacy (SE) directly impacts MOOC continuation (Jung & Lee, 2018; Wu & Chen, 2017). It is necessary to review and summarise research and explore the framework of CI in MOOCs to offer insights into overcoming these issues through its influencing factors. Second, the impact of whether the course takes strict measures to control participation has not been sufficiently studied. IS research attributes mandated and voluntary use to two separate information technology (IT) usage environments (Du et al., 2022). According to mandatory participation, MOOC types are divided into mandatory and self-paced extracurricular tasks. A mandatory course in MOOCs is an online course arranged by the school, which requires students to attend and participate in assessments to obtain college credits. Correspondingly, a self-paced course in MOOCs refers to learners’ active learning of an online course based on their own will, which is a non-mandatory out-of-classroom task. According to the theory of planned behaviour, perceived social pressures (or subjective norms) may make learners carry out a behaviour (or not) (Cheon et al., 2012). However, M. Zhou (2016) suggested that external pressure or demand did not interfere with students’ decisions to opt in or out of a MOOC. Due to the unsystematic study of MOOC course types, it is necessary to deeply study learners’ behaviour in relation to different subdivided course types and explore the influencing factors of learners' CI of different MOOC types to provide targeted insights for personalised services. Third, the role of prior learning experience in mediating MOOC CI has not attracted much attention. The transition from traditional classroom learning to online learning cannot be accomplished overnight, because users need time to adjust (Arbaugh, 2004), and users’ beliefs and attitudes change over time (Venkatesh, 2000). Prior learning experience helps students actively participate in learning (Milligan et al., 2013), and the more experienced learners are, the more rational they are (Wang et al., 2019). However, with the increase in online learning time, positive as well as adverse experiences will increase simultaneously. Negative events affect learner SA (K.-L. Lin et al., 2011), thereby affecting CI. Due to the lack of in-depth analyses of learning experiences, it is necessary to pay more attention to learning behaviour in the context of these experiences. Based on the above discussion, two research questions were proposed as follows: • Given the contradictory conclusions presented in single studies focusing on specific situations and when multiple studies focusing on different situations are regarded as a class of scientific problems, what integrated conclusions can be drawn about the influence of these key factors on MOOC learners’ CI? • How did mandatory participation and learners’ prior learning experience moderate their CI? To answer the first question, this study attempted to use meta-analysis to provide a comprehensive theoretical framework for MOOC learners’ CI. The meta-analysis, proposed by Glass (1976), enables statistical analysis of quantitative data from different studies to integrate research results, which draws accurate and credible conclusions and is applied to identify critical factors. Therefore, this study first constructed a visual knowledge map of relationships among factors in existing research based on the selected literature and then obtained a comprehensive model of MOOC learners’ CI through meta- analysis. To answer the second question, subgroup analysis was applied to classify the study sample types corresponding to two moderator variables, moderate them to establish cumulative knowledge and provide more accurate and robust action guidelines for practice (Ringquist, 2013). Australasian Journal of Educational Technology, 2023, 39(2). 117 Background Factors influencing IS users’ CI CI of IS focuses on influencing factors of the long-term use of ISs (Franque et al., 2020). The TAM and the ECM are the most common models explaining CI in ISs. The TAM is the most widely recognised behavioural intention model in IS disciplines (Q. Ma & Liu, 2004) and is used to examine the relationship between PU, perceived ease of use (PEOU), and attitude (ATT). PU is the degree to which a particular system is perceived to improve performance (Sánchez & Hueros, 2010). PEOU is the degree to which people think using a particular system would be reasonably straightforward (Saadé & Bahli, 2005). ATT refers to how favourable or unfavourable someone evaluates or appraises the behaviours in question (Fishbein & Ajzen, 1977). However, the TAM concerns only short- term beliefs and attitudes before or after the acceptance of an IS and has limitations in explaining eventual CI (Joo et al., 2018). Thus, the TAM has better explanatory power when combined with more external factors (C. Lin et al., 2012). Compared with the TAM, the ECM focuses on factors affecting CI (Bhattacherjee, 2001), which theorises evaluations of system usage based on past experiences. The ECM introduces SAT to explain CI. SAT reflects the positive or pleasant emotional state in work evaluation (Locke, 1976), which is easily affected by confirmation of previous PU and IS usage expectations. Confirmation (CON) refers to the agreement between users’ expectations and their performance in the IS (Bhattacherjee, 2001), which is the main factor affecting user SAT (Hu & Zhang, 2016). Factors influencing MOOC learners’ CI The TAM and ECM have been extensively tested to effectively interpret the CI of online education (Daneji et al., 2019; Fang et al., 2019). Learners' PU, PEOU, ATT, CON and SAT of MOOCs have been found to significantly impact CI in MOOCs (Dai et al., 2020; Daneji et al., 2019; Shao, 2018). Inevitably, as this body of work has grown, the empirical results are scattered and contradictory in some cases, generating puzzles and obstacles to theoretical research and practical application. Exploring the influencing factors of MOOC learners’ CI from multiple perspectives is necessary. In general, social environmental, course-related and learner-related factors are the three most significant factors in studying MOOC learners’ behaviour. Social environmental factors can play a significant role in users’ IS and IT adoption behaviour (Venkatesh et al., 2003). Social presence (SP) describes the salience of others and interpersonal relationships in interaction (Short et al., 1976), which is necessary to ensure an effective online learning environment (Garrison et al., 1999). Social influence (SI) in MOOCs refers to learners believing that important people would like them to continue learning through MOOCs (Venkatesh et al., 2003). Some MOOC learners, especially new learners, are very concerned about media coverage and advice from the people around them (J. Zhou, 2017). From the perspective of the course, since the epidemic, MOOCs should meet the teaching needs brought about by the suspension of face-to-face teaching in schools, as well as the functional learning needs of learners. Task-technology fit (TTF) predicts learner performance and CI in MOOCs (Kim & Song, 2022; W.- S. Lin, 2012). TTF refers to a user’s subjective assessment of whether a technology assists their tasks (Goodhue & Thompson, 1995). Including learner-related factors in MOOCs is also significant (L. Ma & Lee, 2019). Perceived enjoyment (PE) refers to the degree to which people find delight without external reinforcement when carrying out missions (Davis et al., 1992). MOOCs provide users with a valuable platform for engaging in learning and serve the hedonic purpose of creating an enjoyable learning experience (Tao et al., 2022). Outcome expectation (OE) means the perceived results of certain actions (Hsu & Chiu, 2004). Outcomes of e- Australasian Journal of Educational Technology, 2023, 39(2). 118 learning include skills-based outcomes, cognitive outcomes and affective outcomes (Yu et al., 2010). SE in social cognitive theory refers to one’s confidence in organising and performing a task to achieve expected goals (Bandura, 2005). Learners with higher SE in MOOCs motivate themselves and regulate their learning for success (Komarraju & Nadler, 2013). The moderating effect of mandatory participation and prior learning experience It is worth noting that little is known about the relative importance of the various predictors of learners’ CI in MOOCs because the results differ across studies and research contexts. As mentioned above, the impact of course type on learners’ CI is evident. In particular, whether the course is mandatory or not has a huge impact. When a MOOC is taken as a mandatory task, it is often regarded as part of the formal learning process and matches the compulsory credit plan. When MOOCs are self- paced extracurricular tasks, students struggle in an open, self-paced learning environment, often leading to low completion rates and procrastination (Dreisiebner et al., 2020). However, the results of a study conducted by Gregori et al. (2018), who believed that the quality of students’ participation would be higher in self-paced extracurricular tasks because of their interests, were different. User experience is an important moderating factor of IS use behaviour (Bhattacherjee & Premkumar, 2004). They found that the duration of learning has the most critical impact on the learning experience in the study of online learning continuation. The index of learning time may be used to reflect the learning experience (K.-M. Lin et al., 2011). Judgements of prior learning experiences are based on the length of time spent using MOOCs. In this study, we divided MOOC learners into a high-experience group and a low-experience group: the former had more than half a year of online learning experience or had participated in at least 1-semester online courses, while the latter refers to those who have participated in MOOCs for less than half a year or a semester. There have been relatively few studies in the literature specifically on low-experience learners. Thus, we classified studies that did not indicate the respondents’ experience into the low-experience group. Identification of factor and relationship network Based on the analysis of the 53 empirical studies in the collected articles, this paper used Gephi to visually present 170 pairs of structural relationships. The visual knowledge map was developed as shown in Figure 1, where the node size increases with the number of times used in the literature. The label of the node that only appeared once was hidden. The maximum structures are in the centre of the figure, namely SAT, ATT, PU, PEOU, CON, SP, SI, PE, OE, SE, TTF and others. It should be noted that some variables may have different naming methods in different studies. Thus, these names were combined and presented as the same variable and a unified concept was used in this study. Moreover, some variables lacking direct theoretical support or discussed in only a few empirical articles were abandoned. Australasian Journal of Educational Technology, 2023, 39(2). 119 Figure 1. Factor and relationship network Model construction The most frequently occurring structures were incorporated into the research framework using the visual knowledge graph tool to construct the MOOC CI research model, as shown in Figure 2. Figure 2. The research model of MOOC learners’ CI Australasian Journal of Educational Technology, 2023, 39(2). 120 Methodology Meta-analysis was applied to determine the influence of critical factors, and group analysis was conducted to determine the moderating effect of moderators. Previous literature has reported standard procedures of meta-analysis. This study followed the steps suggested by Lipsey and Wilson (2001) and Sabherwal et al. (2006), consisting of (a) searching for individual studies in the literature, (b) coding the identified studies and (c) analysing the accumulated findings. The steps are explained in more detail below. Search process and eligibility criteria The term MOOC was first created to describe the Connectivism and Connected Knowledge online course run by the University of Manitoba in 2008 (Goldie, 2016). Therefore, this study set the period for selecting the research samples from 2008 to 2022. First, the following sites were visited: EBSCO, Web of Science, Elsevier, Emerald, Academic Search Complete, IEEE Xplore, ProQuest and Google Scholar to retrieve related studies by using the search formula of “(MOOC OR MOOCs OR e-learning OR online learning) AND (continuance intention OR continuous usage OR continuation)”. Unpublished conference papers and proceedings were sought from conference websites, such as the Association for Educational Communications and Technology, the Association for the Advancement of Computing in Education, Sloan-C and the American Educational Research Association. This initial retrieval process identified 1826 studies suitable for meta-analysis. Then, three experts in the field of online education research were invited to identify those works of literature highly related to the topic of MOOC CI by screening the titles and abstracts. Through this process, 236 studies were selected from the initial papers. Moreover, we also reviewed the references of the confirmed highly related literature and conducted a forward literature search. Those that were highly relevant to this study but not yet included in the research sample pool were identified and supplemented into the study sample through manual search, which resulted in an additional 10 papers added to the literature search results. Among the 246 articles, 114 articles were empirical studies. To ensure that all the data samples were suitable for the meta-analysis with predetermined criteria, we carefully read the abstract and appraised the research content of each article. Three screening conditions were adopted to clean the data samples: (a) must be an empirical study that investigated the learners’ MOOC CI; (b) must report quantitative information about variables, including sample sizes, correlation coefficients, or other statistical data, such as t values, regression coefficients, mean and standard deviation; and (c) investigated subject must be an individual MOOC learner. Figure 3 illustrates the information and selection process of the included studies from published papers. Finally, 52 articles (with 53 studies) satisfied the above requirements and could be used for this meta-analysis because some articles reported more than one study. Studies of different sample sizes and correlation coefficients in the same article were considered different. See Appendices A and B for detailed information on these studies. Australasian Journal of Educational Technology, 2023, 39(2). 121 Figure 3. Flowchart of literature search Coding of the studies In the process of encoding the literature, we collected the following information: the authors’ name, year of publication, title, sample size, type of online learning platform, learning experience, research model or theory, conceptual model, the key variables included and their Cronbach’s alpha coefficient or comprehensive reliability value and the effect size. In addition, similar variables were merged into one variable. For instance, both perceived hedonic value and perceived fun meant learners’ PE. Based on the research model, this study conducted descriptive statistics on the collected path relationships in the 53 studies, obtained their correlation coefficients and reliability and tested the consistency and stability of the studies. The correlation coefficient r was used as the magnitude of influence in the meta-analysis. The regression coefficient can be used for studies where the correlation coefficient is not given (Wolf, 1986). Some studies took the form of structural equation models, and the following formula was used to convert the t value of a path into a correlation coefficient (Fleiss, 1993): r = #𝑡!/(𝑡! + 𝑑𝑓) (1) where t represents the t value of the path and df represents the degrees of freedom. Analysis of the accumulated findings In the meta-analysis, to avoid measurement errors and sampling errors in different studies, the basic calculation formula in Excel was used to calculate the simple average of the correlation coefficient of each pair of relationships and calculated the adjusted average based on the sample size according to the following formula: 𝑟" = 𝛴𝑁#𝑟# 𝛴𝑁#⁄ (2) where ri is the correlation in each Study i, and Ni is the number of samples. Australasian Journal of Educational Technology, 2023, 39(2). 122 Then, the Fisher r to z transformation on the path correlation coefficient was performed to obtain the changed correlation coefficient z as the merging effect value (Wolf, 1986) to adjust the data deviation caused by sampling variances. The formulas are as follows: 𝑧 = 0.5 ∗ 𝑙𝑛((1 + 𝑟) (1 − 𝑟)⁄ ) (3) 𝑧" = 𝛴𝑁#𝑍# 𝛴𝑁𝑖⁄ (4) 𝑟$ = 𝐸𝑥𝑝(2𝑧" − 1) 𝐸𝑥𝑝(2𝑧" + 1)⁄ (5) where r is the relative correlation and Ni represents the sample size of Study i. We applied the meta-analysis R software package to complete the correlation analysis in this study. The Metacor and Metabias functions were used to calculate the effect size, with a confidence interval of 95%, z test, heterogeneous statistics Q value, heterogeneous index I! and Egger’s test. The 95% confidence interval was calculated to interpret the significance of the average effect size, and the 95% confidence interval excluding 0 suggests that the mean effect size is significant. The z test was used to evaluate the significance of the effect size of the relationship (Cram et al., 2019). The heterogeneity test was used to select a random-effects model or fixed-effects model for meta-analysis. The estimated average effect under the fixed-effects model was often more conservative than that under the random- effects model (Poole & Greenland, 1999). Therefore, the random-effects model was chosen as a theoretical method for the synthesis (Hedges & Vevea, 1998). Egger’s test was used to test publication bias; if p > 0.05, the sample study had no publication bias and was regarded as reliable (Egger et al., 1997). Finally, a subgroup analysis (Q test) based on uniformity estimation was used to discover potential moderating effects. Results Descriptive statistics Table 1 introduces the path relationship of the variables in the research model and their statistical data, including sample size and correlation coefficient. This study examined 17 pairs of relationships. Among them, SAT-CI was detected the most, with 29 studies, followed by PU-CI (28 studies), while PU-ATT and TTF-CI were detected the least, with only 5 studies. In most of these studies, the significance level was higher than 80%, and the average sample size of the path relationship was more significant than 200. Australasian Journal of Educational Technology, 2023, 39(2). 123 Table 1 Descriptive statistics Path relationship No. of studies Correlations Range of correlations Range of sample sizes Average sample size Cumulative sample size Significant Nonsignificant Significant (%) Lower Upper Lower Upper SE-CI 7 7 0 100% 0.17 0.43 144 397 260.86 1826 PE-CI 8 7 1 87.50% 0.00 0.58 126 456 278.5 2228 SP-CI 7 5 2 71.43% -0.21 0.44 144 456 315.29 2207 SAT-CI 29 28 1 96.55% 0.04 0.92 48 1347 370.75 10381 PEOU-CI 9 5 4 55.56% -0.179 0.33 151 456 255.67 2301 PU-CI 28 25 3 89.29% 0.07 0.74 88 1347 375.04 10126 CON-SAT 20 20 0 100% 0.18 0.91 88 1347 449.37 8538 PU-SAT 18 16 2 88.89% 0.04 0.74 88 1347 393.88 6696 PEOU-PU 10 10 0 100% 0.25 0.64 135 2530 467.7 4677 CON-PU 14 14 0 100% 0.17 0.93 88 1347 438.31 5698 OE-CI 7 7 0 100% 0.12 0.50 240 854 389.29 2725 ATT-CI 11 11 0 100% 0.16 0.91 94 2530 527.27 5800 SI-CI 6 5 1 83.33% -0.06 0.41 151 435 256.17 1537 PEOU-ATT 6 5 1 83.33% 0.02 0.23 135 2530 644.5 3867 PU-ATT 5 5 0 100% 0.18 0.72 230 2530 746 3732 TTF-CI 5 4 1 80% 0.11 0.35 252 854 469.8 2349 TTF-PU 6 6 0 100% 0.163 0.742 88 854 345.67 2074 Australasian Journal of Educational Technology, 2023, 39(2). 124 Reliability statistics This study also checked and collected Cronbach’s alpha or comprehensive reliability values (if Cronbach’s alpha is not reported in the literature) to ensure that these variables achieved the desired reliability. As shown in Table 2, the average reliability coefficients of the 12 variables ranged from 0.87 to 0.91, exceeding the recommended threshold of 0.7 (Nunnally, 1994). All variables met the requirements and could be used in the research. Table 2 Reliability statistics Variable Average Minimum Maximum Variance No. of studies PU 0.89 0.60 0.99 0.004 33 PEOU 0.88 0.71 0.99 0.004 16 SE 0.89 0.84 0.93 0.001 7 PE 0.89 0.82 0.95 0.002 8 OE 0.88 0.76 1.00 0.007 7 CON 0.88 0.80 0.94 0.002 19 SP 0.91 0.84 0.95 0.002 5 SAT 0.89 0.65 0.97 0.004 31 CI 0.88 0.71 0.97 0.004 50 ATT 0.90 0.83 0.96 0.001 11 SI 0.87 0.67 0.97 0.012 6 TTF 0.90 0.79 0.98 0.003 9 Correlation analysis Table 3 shows the simple average r of the correlation coefficient, the weighted average r" of the correlation coefficient, the effect size r% after Fisher r to z transformation and its standard error SE, 95% confidence interval, Q statistics and their p values, the z score and p value of the z test, I! statistics and p values of the Egger’s test. Among them, the z test is used to evaluate the importance of the impact size of each relationship (Cram et al., 2019). The z test results showed that at the p < 0.05 level, the effect size of each relationship was statistically significant. The use of the Q value and I! for heterogeneity testing helped in choosing a random-effects model or a fixed-effects model for meta-analysis. All relationships were significant for the heterogeneity test, with Q > K-1, where K was the number of corresponding studies, P(') < 0.05, and I! > 60% (Higgins & Thompson, 2002). Thus, the random-effects model was chosen for this study’s meta-analysis. Australasian Journal of Educational Technology, 2023, 39(2). 125 Table 3 Correlation analysis Path r r+ rz SE 95% Confidence interval z test P(Z) Q value P(Q) I2（%） P(Egger’s test) SE-CI 0.30 0.29 0.30 0.04 0.21–0.38 6.52 0.000 23.40 0.001 74.4 0.72 PE-CI 0.27 0.23 0.28 0.07 0.14–0.42 3.74 0.000 100.11 0.000 93.0 0.67 SP-CI 0.07 0.08 0.07 0.08 -0.09-0.23 0.87 0.386 77.77 0.000 92.3 0.36 SAT-CI 0.47 0.50 0.51 0.04 0.41–0.60 8.91 0.000 1009.35 0.000 97.2 0.28 PEOU-PU 0.39 0.34 0.40 0.04 0.32–0.48 8.66 0.000 67.06 0.000 86.6 0.07 PEOU-CI 0.10 0.10 0.11 0.05 0.00–0.21 2.04 0.042 49.95 0.000 84.0 0.71 PU-CI 0.29 0.31 0.30 0.04 0.23–0.38 7.38 0.000 392.58 0.000 93.1 0.57 CON-SAT 0.46 0.53 0.50 0.05 0.38–0.61 7.02 0.000 1010.32 0.000 98.1 0.04 PU-SAT 0.36 0.28 0.38 0.05 0.27–0.48 6.50 0.000 372.96 0.000 95.4 0.00 CON-PU 0.49 0.47 0.55 0.06 0.38–0.68 5.63 0.000 1084.31 0.000 98.8 0.60 OE-CI 0.23 0.23 0.24 0.06 0.12–0.35 3.89 0.000 68.78 0.000 91.3 0.84 ATT-CI 0.58 0.67 0.65 0.08 0.45–0.79 5.26 0.000 1276.74 0.000 99.2 0.12 SI-CI 0.21 0.23 0.22 0.08 0.06–0.36 2.70 0.007 55.11 0.000 90.9 0.40 PEOU-ATT 0.16 0.15 0.16 0.03 0.11–0.21 5.99 0.000 8.44 0.134 40.7 0.68 PU-ATT 0.48 0.59 0.51 0.10 0.29–0.68 4.17 0.000 165.47 0.000 97.6 0.25 TTF-PU 0.45 0.45 0.47 0.08 0.30–0.62 4.90 0.000 78.87 0.000 93.7 0.86 TTF-CI 0.18 0.17 0.18 0.05 0.09–0.27 3.82 0.000 20.42 0.000 80.4 0.40 Note. P(') is the significance level of the Q test for heterogeneity; P(%) is the significance level of the z test. Australasian Journal of Educational Technology, 2023, 39(2). 126 According to the heterogeneity test results, except for the PEOU-ATT path, the heterogeneity of the other paths’ effect sizes was significant. Therefore, the fixed-effects model was selected to analyse the PEOU- ATT relationship. Forest plots are usually employed to visualise heterogeneous test results, and the results are shown in Figure 4. The PEOU-ATT effect value is 0.15, and the confidence interval is 0.12–0.18. Figure 4. The forest plot of PEOU-ATT The forest plots can visualise the heterogeneous test results. Figure 5 shows the range of R(z), central tendency and correlation coefficient in the random-effects model. Most of the relationships pass the significance test. However, the SP-CI path includes 0 in the 95% confidence interval, and the p value of the z test exceeds 0.05, which indicates that the effect size is not significant. Regarding publication bias, when Egger’s test value exceeds 0.05, it indicates that the sample study has no publication bias (Egger et al., 1997). The PU-SAT relationship failed the Egger’s test, for P (Egger’s test) = 0.00. Cohen (2013) pointed out that an effect size close to 0.1 means that the effect on the dependent variable is small. An effect size close to 0.3 indicates a medium effect, while an effect size close to 0.5 indicates a relatively high effect (Cohen, 2013). For central tendency, PU-CI, SE-CI, PEOU-PU and TTF-CI are more concentrated than other relationships. According to the correlation analysis, the effect value of the ATT- CI relationship is the largest at 0.65, revealing that ATT has the most substantial explanatory power for MOOC CI. The effect size of PU-ATT, CON-PU and CON-SAT is also greater than 0.5. The effect size of SE- CI, PEOU-PU, PU-CI, PU-SAT and TTF-PU is between 0.3 and 0.5, indicating strong effects. The effect size of PE-CI, OE-CI and SI-CI is between 0.2 and 0.3, which means a medium effect. The effect size of PEOU-CI is close to 0.1, indicating a low effect. Figure 5. The forest plots of R(z) Australasian Journal of Educational Technology, 2023, 39(2). 127 Moderator analysis Table 4 shows the results of the moderator analysis. Concerning the moderating effect of mandatory participation, the PEOU-CI relationship is moderated by MOOC mandatory participation. The effect of the self-paced extracurricular tasks (r% = 0.23) is greater than the mandatory tasks (r% = 0.41). Concerning the moderating effect of prior learning experience, the PU-CI and CON-SAT relationships are influenced by prior learning experience. For the PU-CI relationship, the low-experience group’s impact size (r% = 0.37) is higher than the impact size of the high-experience group (r% = 0.20). For the CON-SAT relationship, the low-experience group’s impact size (r% = 0.60) is higher than the impact size of the high-experience group (r% = 0.37). Figures 6 (a), (b) and (c) visualise the overall distribution of variable values and feature values through violin plots. Australasian Journal of Educational Technology, 2023, 39(2). 128 Table 4 Brief results of moderator analysis Path relationship Subgroups k N r% 95% CI Q* Between groups test Q+ p Moderator 1: MOOC mandatory participation PEOU-CI Self-paced 3 692 0.23 0.07-0.38 10.34** 3.76* 0.052 Mandatory 6 1609 0.41 -0.06-0.14 21.03*** Moderator 2: Prior learning experience PU-CI Less 17 7163 0.37 0.26-0.45 237.18*** 4.39** 0.036 More 12 3217 0.20 0.10-0.29 79.31*** CON-SAT Less 11 6007 0.60 0.43-0.72 572.45*** 4.26** 0.039 More 8 2421 0.37 0.23-0.50 122.68*** Note. k is the number of studies; N is the number of observations in each study; r% means correlation; Q* is the Q test for homogeneity within subgroups; Q+ is the Q test for homogeneity between subgroups; p is the significance level of the Q test for heterogeneity between subgroups. *p < 0.1. **p < 0.05. ***p < 0.01. Australasian Journal of Educational Technology, 2023, 39(1). 129 (a) (b) (c) Figure 6. (a) Violin plots of PEOU-CI moderated by mandatory participation; (b) violin plots of PU-CI moderated by prior learning experience; (c) violin plots of CON-SAT moderated by prior learning experience Discussion The objective of this study was to clarify the relative importance of the critical factors to learners’ MOOC CI, as well as the moderating effect of specific moderators. The final model is shown in Figure 7. According to the meta-analysis results, traditional CI theories are valid in MOOC CI. This study further confirmed the stability of the TAM and ECM models in research on MOOC CI. ATT and SAT were crucial ways to determine and explain MOOC CI, which is consistent with prior findings (Alraimi et al., 2015; Wu & Chen, 2017). The findings confirmed that CON positively impacted the perceived performance of MOOC platforms, including increased PU and SAT (Gu et al., 2021), which in turn significantly impacted MOOC CI (Alraimi et al., 2015). The path of PU-ATT is also significant. PEOU affected CI in MOOCs directly and indirectly affected CI through PU and ATT, which is consistent with research (Shao, 2018). While there is a publication bias towards PU-SAT, indicating that PU provides limited support for improving MOOC SAT (Alraimi et al., 2015; Daneji et al., 2019). MOOC learners usually care about their personal needs when using MOOC platforms (Olasina, 2018). Regarding social environmental factors as a new learning experience, SI positively influences learners’ MOOC CI. That is, users’ MOOC CI is influenced by word of mouth from the media and those around them. In comparison, the path effect of SP-CI is not significant and needs further study. Some studies have concluded that study group members’ SP plays a vital role in driving learners’ MOOC CI (Dalvi-Esfahani et al., 2020; Luo et al., 2018). Other studies have shown that online interactions negatively influence CI with regard to participation (Chang et al., 2015; Zhang et al., 2012). For course-related factors, TTF is significantly positively correlated with users' PU and MOOC CI (Kim & Song, 2022). Meanwhile, the indirect effect of TTF on MOOC CI is more significant than the direct effect because the effect size of TTF-PU is more significant than that of TTF-CI. That is, assessing the relationship between TTF and MOOC CI will likely highlight other mediators, such as PU, in more detail. Regarding learner-related factors, SE, PE and Australasian Journal of Educational Technology, 2023, 39(1). 130 OE positively affect MOOC CI. It is essential to properly measure and enhance learners’ SE in MOOCs (Lee et al., 2020). PE is emotional arousal. Students may view MOOCs as hedonic systems, in addition to using the platform as a utilitarian system for learning (Tao et al., 2022). For OE, MOOC platforms and teachers should strive to meet learner expectations, enhancing MOOC CI (Bourdeaux & Schoenack, 2016). Figure 7. The revised research model of MOOC CI Note. The significance level is depicted as a solid line, and the nonsignificant line is portrayed as a dotted line. The moderating effect of mandatory participation and prior learning experience was also confirmed. For the moderating effect of mandatory participation, PEOU plays a more significant role in mandatory tasks than in self-paced tasks. This finding suggests that learners in mandatory tasks may have higher requirements for system perception because they must use it, so they pay more attention to PEOU, while students in self-paced tasks may pay more attention to other factors. For the moderating effect of prior learning experience, the impact of PU-CI in the low-experience group is greater than that in the high- experience group. When the MOOC platform improves the performance of beginners and makes them perceive the usefulness of MOOCs, learners will have positive psychological feedback on the platform, thereby enhancing CI in MOOCs (Gu et al., 2021). In addition, for CON-SAT, the confirmation of the platform by learners with less learning experience has a more significant effect on their SAT. Confirmation changes as the user’s experience with a particular technology increases (Chauhan et al., 2022). Practical implications The findings are of practical value to MOOC developers, especially when face-to-face teaching is greatly affected by the COVID-19 epidemic. First, the content quality of the MOOC platform should be improved to match the content needs of students. In designing and promoting MOOC course content, platforms should provide standard and scientific training courses for course developers to help them develop a higher-quality course content system. Subject teachers can upload course-related materials in a targeted manner to facilitate students’ learning at different paces. Only in this way can learners’ ATT and SA be significantly enhanced. In addition, big data technology can also be used to track individual learning traces and group learning transfer, accurately judge and develop new curriculum systems and eliminate unpopular curriculum systems. Meanwhile, the platforms must be carefully publicised to avoid exaggerating benefits and system costs because recognition is closely related to SAT. Australasian Journal of Educational Technology, 2023, 39(1). 131 Second, MOOC developers should pay enough attention to improving the platform’s social quality to match students’ social needs. On the one hand, practitioners should attract more universities and more high-level teachers who are deeply involved in a specific subject domain. Industry experts should provide high-quality courses to enhance the external SI and reputation of MOOC platforms (J. Zhou, 2017). On the other hand, an exciting learning environment should be created by implementing advanced technology (Guo et al., 2016) or gaming elements to improve learners’ internal social participation and PE. For instance, gamification elements, such as points, badges or rewards, are displayed on leaderboards and provide a sense of competition between learners and their online classmates (Rohan et al., 2021) to motivate learners to be more actively engaged with MOOCs. In addition, positive feedback on learning outcomes should be provided regularly, with assessments of learning through forums, question-and- answer sessions, quizzes and automatic scoring of papers (Xiong & Suen, 2018) to motivate them to achieve learning goals and meet expectations for outcomes. Finally, MOOC platforms should strive to meet the individual service needs of learners. As mentioned, personalised, customised services according to platform types and prior experiences are vital. MOOC platforms should establish and maintain close relationships with new learners in MOOCs. MOOC providers can provide convenient feedback channels to follow learners’ perceptions. For example, regular questionnaire surveys should be conducted to listen to students’ real needs and voices and increase their PU. Moreover, SAT can be promoted by ensuring the confirmation of expectations. In addition, mandatory courses in MOOCs should pay special attention to improving learners’ PEOU, enhancing CI and improving teaching efficiency. It is equally important to actively establish instant feedback channels for problem solving through study groups to reduce the use barriers of MOOC platforms. Limitations and future directions This study is a meaningful exploration of MOOC learning behaviour research, but there are still some limitations, which leaves space for further research. The quantitative research methods used in this study have specific requirements for the scale of research samples. Although this study included and analysed critical variables in the literature, some significant factors may not be included in this study because of their relative novelty and rare occurrence. In the future, scholars could conduct further empirical research on those variables and relationships to obtain richer research conclusions. As English is the world’s most popular language for scientific communication, this study included only studies published in English. Studies written in other languages were not included, which may limit the generalisability of the findings. In future studies, publications in multiple languages could be collected to enrich the application scope of the research conclusions. Moreover, with the maturity of MOOC industry development and the enrichment of the MOOC research system, there will be an increasing number of interesting topics, such as comparative studies of different cultural situations and innovative research on the MOOC metaverse, which may encourage new research topics. Author contributions Min Zhang: Conceptualisation, Methodology, Formal analysis, Writing original draft, Writing – review and editing; Sihong Li: Data curation, Analyzing and interpreting the data, Visualization, Writing original draft; Yan Zhang: Conceptualisation, Methodology, Writing part of the original draft, Writing – review and editing, revising. Acknowledgements This research is supported by the National Natural Science Foundation of China (Grant Number: 72074174), the Humanity and Social Science Foundation of Ministry of Education of China (Grant Number: 18YJA870016), the Fundamental Research Funds for the Central Universities, and the Fundamental Research Funds for the Central Universities “Research on the potential risk of the meta universe and its legal governance” (Grant Number: E2E42109X2). Australasian Journal of Educational Technology, 2023, 39(1). 132 References (References marked with an asterisk indicate studies included in the meta-analysis.) Alemayehu, L., & Chen, H. (2021). Learner and instructor-related challenges for learners’ engagement in MOOCs: A review of 2014–2020 publications in selected SSCI indexed journals. Interactive Learning Environments, 1–23. https://doi.org/10.1080/10494820.2021.1920430 *Alraimi, K. M., Zo, H., & Ciganek, A. P. (2015). Understanding the MOOCs continuance: The role of openness and reputation. Computers & Education, 80, 28–38. https://doi.org/10.1016/j.compedu.2014.08.006 Arbaugh, J. (2004). Learning to learn online: A study of perceptual changes between multiple online course experiences. The Internet and Higher Education, 7(3), 169–182. https://doi.org/10.1016/j.iheduc.2004.06.001 Bandura, A. (2005). The primacy of self-regulation in health promotion. Applied Psychology, 54(2), 245– 254. https://doi.org/10.1111/j.1464-0597.2005.00208.x Bhattacherjee, A. (2001). Understanding information systems continuance: An expectation-confirmation model. MIS Quarterly, 25(3), 351–370. https://doi.org/10.2307/3250921 Bhattacherjee, A., & Premkumar, G. (2004). Understanding changes in belief and attitude toward information technology usage: A theoretical model and longitudinal test. MIS Quarterly, 28(2), 229– 254. https://doi.org/10.2307/25148634 Bourdeaux, R., & Schoenack, L. (2016). Adult student expectations and experiences in an online learning environment. The Journal of Continuing Higher Education, 64(3), 152–161. https://doi.org/10.1080/07377363.2016.1229072 *Brahmasrene, T., & Lee, J. W. (2012). Determinants of intent to vontinue using online learning: A tale of two universities. Interdisciplinary Journal of Information, Knowledge, and Management, 7(1), 1– 20. https://doi.org/10.28945/1548 *Chang, C.-C., Liang, C., Shu, K., & Chiu, Y. (2015). Alteration of influencing factors of e-learning continued intention for different degrees of online participation. The International Review of Research in Open and Distributed Learning, 16(4), 33–61. https://doi.org/10.19173/irrodl.v16i4.2084 *Chauhan, S., Goyal, S., Bhardwaj, A. K., & Sergi, B. S. (2022). Examining continuance intention in business schools with digital classroom methods during COVID-19: A comparative study of India and Italy. Behaviour & Information Technology, 41(8), 1596–1619. https://doi.org/10.1080/0144929X.2021.1892191 *Chen, C.-C., Lee, C.-H. and Hsiao, K.-L. (2018). Comparing the determinants of non-MOOC and MOOC continuance intention in Taiwan: Effects of interactivity and openness. Library Hi Tech, 36(4), 705– 719. https://doi.org/10.1108/LHT-11-2016-0129 *Cheng, Y.-M. (2019). How does task-technology fit influence cloud-based e-learning continuance and impact? Education + Training, 61(4), 480–499. https://doi.org/10.1108/ET-09-2018-0203 *Cheng, Y.-M. (2022). How different categories of gamified stimuli affect massive open online courses continuance intention and learning performance? Mediating roles of internal experiences. Social Science Computer Review. https://doi.org/10.1177/08944393221111928 Cheon, J., Lee, S., Crooks, S. M., & Song, J. (2012). An investigation of mobile learning readiness in higher education based on the theory of planned behavior. Computers & Education, 59(3), 1054–1064. https://doi.org/10.1016/j.compedu.2012.04.015 *Chiu, C.-M., & Wang, E. T. (2008). Understanding Web-based learning continuance intention: The role of subjective task value. Information & Management, 45(3), 194–201. https://doi.org/10.1016/j.im.2008.02.003 Cohen, J. (2013). Statistical power analysis for the behavioral sciences. Routledge. Cram, W. A., D’arcy, J., & Proudfoot, J. G. (2019). Seeing the forest and the trees: a meta-analysis of the antecedents to information security policy compliance. MIS Quarterly, 43(2), 525–554. https://doi.org/10.25300/MISQ/2019/15117 *Dağhan, G., & Akkoyunlu, B. (2016). Modeling the continuance usage intention of online learning environments. Computers in Human Behavior, 60, 198–211. https://doi.org/10.1016/j.chb.2016.02.066 Australasian Journal of Educational Technology, 2023, 39(1). 133 *Dai, H. M., Teo, T., & Rappa, N. A. (2020). Understanding continuance intention among MOOC participants: The role of habit and MOOC performance. Computers in Human Behavior, 112, Article 106455. https://doi.org/10.1016/j.chb.2020.106455 *Dai, H. M., Teo, T., & Rappa, N. A. (2022). The role of gender and employment status in MOOC learning: An exploratory study. Journal of Computer Assisted Learning, 38(5), 1360–1370. https://doi.org/10.1111/jcal.12681 *Dai, H. M., Teo, T., Rappa, N. A., & Huang, F. (2020). Explaining Chinese university students’ continuance learning intention in the MOOC setting: A modified expectation confirmation model perspective. Computers & education, 150, 103850. https://doi.org/10.1016/j.compedu.2020.103850 *Dalvi-Esfahani, M., Wai Leong, L., Ibrahim, O., & Nilashi, M. (2020). Explaining students’ continuance intention to use Mobile Web 2.0 learning and their perceived learning: An integrated approach. Journal of Educational Computing Research, 57(8), 1956–2005. https://doi.org/10.1177/0735633118805211 *Daneji, A. A., Ayub, A. F. M., & Khambari, M. N. M. (2019). The effects of perceived usefulness, confirmation and satisfaction on continuance intention in using massive open online course (MOOC). Knowledge Management & E-Learning, 11(2), 201–214. https://doi.org/10.34105/j.kmel.2019.11.010 Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1992). Extrinsic and intrinsic motivation to use computers in the workplace. Journal of Applied Social Psychology, 22(14), 1111–1132. https://doi.org/10.1111/j.1559-1816.1992.tb00945.x *de Melo Pereira, F. A., Ramos, A. S. M., Gouvêa, M. A., & da Costa, M. F. (2015). Satisfaction and continuous use intention of e-learning service in Brazilian public organizations. Computers in Human Behavior, 46, 139–148. https://doi.org/10.1016/j.chb.2015.01.016 Dreisiebner, S., Polzer, A. K., Robinson, L., Libbrecht, P., Boté-Vericad, J.-J., Urbano, C., Mandl, T., Vilar, P., Žumer, M., & Juric, M. (2020). Facilitation of information literacy through a multilingual MOOC considering cultural aspects. Journal of Documentation, 77(3), 777–797. https://doi.org/10.1108/JD- 06-2020-0099 Du, Z., Wang, F., Wang, S., & Xiao, X. (2022). Enhancing learner participation in online discussion forums in massive open online courses: The role of mandatory participation. Frontiers in Psychology, 13. https://doi.org/10.3389/fpsyg.2022.819640 Egger, M., Smith, G. D., Schneider, M., & Minder, C. (1997). Bias in meta-analysis detected by a simple, graphical test. BMJ, 315(7109), 629–634. https://doi.org/10.1136/bmj.315.7109.629 Fang, J., Tang, L., Yang, J., & Peng, M. (2019). Social interaction in MOOCs: The mediating effects of immersive experience and psychological needs satisfaction. Telematics and Informatics, 39, 75–91. https://doi.org/10.1016/j.tele.2019.01.006 Fishbein, M., & Ajzen, I. (1977). Belief, attitude, intention, and behavior: An introduction to theory and research. Philosophy & Rhetoric, 10(2), 130–132. https://philarchive.org/archive/FISBAI Fleiss, J. (1993). Review papers: The statistical basis of meta-analysis. Statistical Methods in Medical Research, 2(2), 121–145. https://doi.org/10.1177/096228029300200202 Franque, F. B., Oliveira, T., Tam, C., & de Oliveira Santini, F. (2020). A meta-analysis of the quantitative studies in continuance intention to use an information system. Internet Research, 31(1), 1–36. https://doi.org/10.1108/INTR-03-2019-0103 Garrison, D. R., Anderson, T., & Archer, W. (1999). Critical inquiry in a text-based environment: Computer conferencing in higher education. The Internet and Higher Education, 2(2-3), 87–105. https://doi.org/10.1016/S1096-7516(00)00016-6 Glass, G. V. (1976). Primary, secondary, and meta-analysis of research. Educational Researcher, 5(10), 3– 8. https://doi.org/10.3102/0013189X005010003 Goldie, J. G. S. (2016). Connectivism: A knowledge learning theory for the digital age? Medical Teacher, 38(10), 1064–1069. https://doi.org/10.3109/0142159X.2016.1173661 Goodhue, D. L., & Thompson, R. L. (1995). Task-technology fit and individual performance. MIS Quarterly, 19, 213–236. https://doi.org/10.2307/249689 Gregori, E. B., Zhang, J., Galván-Fernández, C., & de Asís Fernández-Navarro, F. (2018). Learner support in MOOCs: Identifying variables linked to completion. Computers & Education, 122, 153–168. https://doi.org/10.1016/j.compedu.2018.03.014 Australasian Journal of Educational Technology, 2023, 39(1). 134 *Gu, W., Xu, Y., & Sun, Z. (2021). Does MOOC quality affect users’ continuance intention? Based on an integrated model. Sustainability, 13(22), Article 12536. http://doi.org/10.3390/su132212536 *Guo, Z., Xiao, L., Van Toorn, C., Lai, Y., & Seo, C. (2016). Promoting online learners’ continuance intention: An integrated flow framework. Information & Management, 53(2), 279–295. https://doi.org/10.1016/j.im.2015.10.010 Hedges, L. V., & Vevea, J. L. (1998). Fixed-and random–effects models in meta-analysis. Psychological Methods, 3(4), 486–504. https://doi.org/10.1037/1082-989X.3.4.486 Higgins, J. P., & Thompson, S. G. (2002). Quantifying heterogeneity in a meta-analysis. Statistics in Medicine, 21(11), 1539–1558. https://doi.org/10.1002/sim.1186 *Hsu, J.-Y., Chen, C.-C., & Ting, P.-F. (2018). Understanding MOOC continuance: An empirical examination of social support theory. Interactive Learning Environments, 26(8), 1100–1118. https://doi.org/10.1080/10494820.2018.1446990 Hsu, M. H., & Chiu, C. M. (2004). Predicting electronic service continuance with a decomposed theory of planned behaviour. Behaviour & Information Technology, 23(5), 359–373. https://doi.org/10.1080/01449290410001669969 Hu, J., & Zhang, Y. (2016). Understanding Chinese undergraduates’ continuance intention to use mobile book-reading apps: An integrated model and empirical study. Libri, 66(2), 85–99. https://doi.org/10.1515/libri-2015-0090 *Ishak, K. A., & Malaysia, S. (2020). Exploring the factors affecting the continuance intention of massive open online courses. In M. S. A. Bakar, H. Nordin, I. N. M. Shaharanee, W. H. W. Ishak, & F. M. Yamin (Eds.), Proceedings of the Inspirational Scholar Symposium (pp. 288–297). Pusat Pengajaran Pembelajaran Universiti. http://utlc.uum.edu.my/images/2021/ebook/Proceeding_ISS2019.pdf *Jo, D. (2018). Exploring the determinants of MOOCs continuance intention. KSII Transactions on Internet and Information Systems (TIIS), 12(8), 3992–4005. https://doi.org/10.3837/tiis.2018.08.024 *Joo, Y. J., So, H.-J., & Kim, N. H. (2018). Examination of relationships among students’ self- determination, technology acceptance, satisfaction, and continuance intention to use K-MOOCs. Computers & Education, 122, 260–272. https://doi.org/10.1016/j.compedu.2018.01.003 *Jung, Y., & Lee, J. (2018). Learning engagement and persistence in massive open online courses (MOOCS). Computers & Education, 122, 9–22. https://doi.org/10.1016/j.compedu.2018.02.013 *Kim, R., & Song, H.-D. (2022). Examining the influence of teaching presence and task-technology fit on continuance intention to use MOOCs. The Asia-Pacific Education Researcher, 31(4), 395–408. https://doi.org/10.1007/s40299-021-00581-x Kizilcec, R. F., Piech, C., & Schneider, E. (2013). Deconstructing disengagement: Analyzing learner subpopulations in massive open online courses. In D. Suthers, K. Verbert, E. Duval, & X. Ochoa (Eds.), Proceedings of the 3rd International Conference on Learning Analytics and Knowledge (pp. 170–179). Association for Computing Machinery. https://doi.org/10.1145/2460296.2460330 Komarraju, M., & Nadler, D. (2013). Self-efficacy and academic achievement: Why do implicit beliefs, goals, and effort regulation matter? Learning and Individual Differences, 25, 67–72. https://doi.org/10.1016/j.lindif.2013.01.005 *Lai, I. K. W., & Lai, D.C.F. (2014). Understanding students’ continuance intention toward social networking e-learning. In S. K. S. Cheung, J. Fong, J. Zhang, R. Kwan, & L. F. Kwok (Eds.), Lecture notes in computer science: Vol. 8595. Hybrid learning: Theory and practice (pp. 170–179). Springer International Publishing. https://doi.org/10.1007/978-3-319-08961-4_17 Lee, D., Watson, S. L., & Watson, W. R. (2020). The influence of successful MOOC learners’ self-regulated learning strategies, self-efficacy, and task value on their perceived effectiveness of a massive open online course. The International Review of Research in Open and Distributed Learning, 21(3), 81–98. https://doi.org/10.19173/irrodl.v21i3.4642 *Lee, M. C. (2010). Explaining and predicting users’ continuance intention toward e-learning: An extension of the expectation–confirmation model. Computers & Education, 54(2), 506–516. https://doi.org/10.1016/j.compedu.2009.09.002 Lin, C., Lin, I.-C., & Roan, J. (2012). Barriers to physicians’ adoption of healthcare information technology: An empirical study on multiple hospitals. Journal of Medical Systems, 36(3), 1965–1977. http://doi.org/10.1007/s10916-011-9656-7 Australasian Journal of Educational Technology, 2023, 39(1). 135 *Lin, K.-M. (2011). E-Learning continuance intention: Moderating effects of user e-learning experience. Computers & Education, 56(2), 515–526. https://doi.org/10.1016/j.compedu.2010.09.017 *Lin, K.-M., Chen, N.-S., & Fang, K. (2011). Understanding e-learning continuance intention: A negative critical incidents perspective. Behaviour & Information Technology, 30(1), 77–89. https://doi.org/10.1080/01449291003752948 Lin, W.-S. (2012). Perceived fit and satisfaction on web learning performance: IS continuance intention and task-technology fit perspectives. International Journal of Human-Computer Studies, 70(7), 498– 507. https://doi.org/10.1016/j.ijhcs.2012.01.006 *Lin, W.-S., & Wang, C.-H. (2012). Antecedences to continued intentions of adopting e-learning system in blended learning instruction: A contingency framework based on models of information system success and task-technology fit. Computers & Education, 58(1), 88–99. https://doi.org/10.1016/j.compedu.2011.07.008 Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. SAGE publications, Inc. Locke, E. A. (1976). The nature and causes of job satisfaction. In M. Dunnette (Ed.), Handbook of industrial and organizational psychology (pp. 1297–1350). Rand McNally. *Lu, Y., Wang, B., & Lu, Y. (2019). Understanding key drivers of MOOC satisfaction and continuance intention to use. Journal of Electronic Commerce Research, 20(2), 105–117. http://www.jecr.org/sites/default/files/2019vol20no2_Paper3.pdf *Luo, N., Zhang, Y., & Zhang, M. (2018). Retaining learners by establishing harmonious relationships in e- learning environment. Interactive Learning Environments, 27(1), 118–131. https://doi.org/10.1080/10494820.2018.1506811 Ma, L., & Lee, C. S. (2019). Investigating the adoption of MOOC s: A technology–user–environment perspective. Journal of Computer Assisted Learning, 35(1), 89–98. https://doi.org/10.1111/jcal.12314 Ma, Q., & Liu, L. (2004). The technology acceptance model: A meta-analysis of empirical findings. Journal of Organizational and End User Computing, 16(1), 59–72. http://doi.org/10.4018/joeuc.2004010104 Milligan, C., Littlejohn, A., & Margaryan, A. (2013). Patterns of engagement in connectivist MOOCs. Journal of Online Learning and Teaching, 9(2), 149–159. https://jolt.merlot.org/vol9no2/milligan_0613.htm *Najmul Islam, A. (2011). Understanding continued usage intention in e-learning context. In eFuture: Creating solutions for the individual, organisations and society—Proceedings of the 24th BLED eConference (pp. 546–557). Association for Information Systems. https://aisel.aisnet.org/cgi/viewcontent.cgi?article=1056&context=bled2011 *Nong, Y., Buavaraporn, N., & Punnakitikashem, P. (2022). Exploring the factors influencing users’ satisfaction and continuance intention of MOOCs in China. Kasetsart Journal of Social Sciences, 43(2), 403–408. https://so04.tci-thaijo.org/index.php/kjss/article/view/258499/175125 *Nugroho, M. A., Setyorini, D., & Novitasari, B. T. (2019). The role of satisfaction on perceived value and e-learning usage continuity relationship. Procedia Computer Science, 161, 82–89. https://doi.org/10.1016/j.procs.2019.11.102 Nunnally, J. C. (1994). Psychometric theory 3E. Tata McGraw-Hill Education. Olasina, G. (2018). Factors of best practices of e-learning among undergraduate students. Knowledge Management & E-Learning: An International Journal, 10(3), 265–289. https://doi.org/10.34105/j.kmel.2018.10.016 *Park, T. J., Whang, J., Watts, S., & Han, D.-G. (2022). Key success factors in the continuous use of MOOC education in South Korea. International Journal of Innovation and Learning, 31(2), 137–165. https://doi.org/10.1504/IJIL.2022.120665 Poole, C., & Greenland, S. (1999). Random-effects meta-analyses are not always conservative. American Journal of Epidemiology, 150(5), 469–475. https://doi.org/10.1093/oxfordjournals.aje.a010035 *Qi, D., Zhang, M., & Zhang, Y. (2020). Resource integration, value co-creation and continuance intention in MOOCs learning process. Interactive Learning Environments, 1–13. https://doi.org/10.1080/10494820.2020.1802299 *Ramayah, T., & Lee, J. W. C. (2012). System characteristics, satisfaction and e-learning usage: a structural equation model (SEM). Turkish Online Journal of Educational Technology, 11(2), 196–206. http://www.tojet.net/articles/v11i2/11221.pdf Ringquist, E. (2013). Meta-analysis for public management and policy. John Wiley & Sons. Australasian Journal of Educational Technology, 2023, 39(1). 136 *Rodríguez-Ardura, I., & Meseguer-Artola, A. (2016). What leads people to keep on e-learning? An empirical analysis of users' experiences and their effects on continuance intention. Interactive Learning Environments, 24(6), 1030–1053. https://doi.org/10.1080/10494820.2014.926275 *Rohan, R., Pal, D., Funilkul, S., Chutimaskul, W., & Eamsinvattana, W. (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 Saadé, R., & Bahli, B. (2005). The impact of cognitive absorption on perceived usefulness and perceived ease of use in on-line learning: An extension of the technology acceptance model. Information & management, 42(2), 317–327. https://doi.org/10.1016/j.im.2003.12.013 Sabherwal, R., Jeyaraj, A., & Chowa, C. (2006). Information system success: Individual and organizational determinants. Management Science, 52(12), 1849–1864. https://doi.org/10.1287/mnsc.1060.0583 Sánchez, R. A., & Hueros, A. D. (2010). Motivational factors that influence the acceptance of Moodle using TAM. Computers in Human Behavior, 26(6), 1632–1640. https://doi.org/10.1016/j.chb.2010.06.011 *Shanshan, S., & Wenfei, L. (2022). Understanding the impact of quality elements on MOOCs continuance intention. Education and Information Technologies, 27(8), 10949–10976. https://doi.org/10.1007/s10639-022-11063-y *Shao, Z. (2018). Examining the impact mechanism of social psychological motivations on individuals’ continuance intention of MOOCs: The moderating effect of gender. Internet Research, 28, 232–250. https://doi.org/10.1108/IntR-11-2016-0335 Short, J., Williams, E., & Christie, B. (1976). The social psychology of telecommunications. Wiley. *Suriazdin, S. A., Hidayanto, A. N., Maulida, M., Kurtinus, A. Y., Arrumaisha, H., Aisyah, N., & Pradana, R. P. (2022). Technology attractiveness and its impact on MOOC continuance intention. International Journal of Emerging Technologies in Learning, 17(4). https://doi.org/10.3991/ijet.v17i04.28853 *Tan, M., & Shao, P. (2015). An ECM-ISC based study on learners’ continuance intention toward e- learning. International Journal of Emerging Technologies in Learning, 10(4), 22–27. http://doi.org/10.3991/ijet.v10i4.4543 Tao, D., Fu, P., Wang, Y., Zhang, T., & Qu, X. (2022). Key characteristics in designing massive open online courses (MOOCs) for user acceptance: An application of the extended technology acceptance model. Interactive Learning Environments, 30(5), 882–895. https://doi.org/10.1080/10494820.2019.1695214 *Tawafak, R. M., Romli, A. B., & Arshah, R. B. A. (2018). Continued intention to use UCOM: Four factors for integrating with a technology acceptance model to moderate the satisfaction of learning. IEEE Access, 6, 66481–66498. https://doi.org/10.1109/ACCESS.2018.2877760 *Tsai, Y.-h., Lin, C.-h., Hong, J.-c., & Tai, K.-h. (2018). The effects of metacognition on online learning interest and continuance to learn with MOOCs. Computers & Education, 121, 18–29. https://doi.org/10.1016/j.compedu.2018.02.011 Venkatesh, V. (2000). Determinants of perceived ease of use: Integrating control, intrinsic motivation, and emotion into the technology acceptance model. Information Systems Research, 11(4), 342–365. https://doi.org/10.1287/isre.11.4.342.11872 Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information technology: Toward a unified view. MIS Quarterly, 27(3), 425–478. https://doi.org/10.2307/30036540 *Wang, L.-Y.-K., Lew, S.-L., Lau, S.-H., & Leow, M.-C. (2019). Usability factors predicting continuance of intention to use cloud e-learning application. Heliyon, 5(6), e01788. https://doi.org/10.1016/j.heliyon.2019.e01788 *Wang, T., Lin, C.-L., & Su, 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 Wang, W., Guo, L., & Sun, R. (2019). Rational herd behavior in online learning: Insights from MOOC. Computers in Human Behavior, 92, 660–669. https://doi.org/10.1016/j.chb.2017.10.009 Wolf, F. M. (1986). Meta-analysis: Quantitative methods for research synthesis. Sage Publications. Australasian Journal of Educational Technology, 2023, 39(1). 137 *Wu, B., & Chen, X. (2017). Continuance intention to use MOOCs: Integrating the technology acceptance model (TAM) and task technology fit (TTF) model. Computers in Human Behavior, 67, 221–232. https://doi.org/10.1016/j.chb.2016.10.028 Xiong, Y., & Suen, H. K. (2018). Assessment approaches in massive open online courses: Possibilities, challenges and future directions. International Review of Education, 64(2), 241–263. https://doi.org/10.1007/s11159-018-9710-5 *Xu, N., & Wang, H. (2017). Research on impact factors of user's continuance intention in online education platform. In Proceedings of the 2017 International Conference on Industrial Informatics – Computing Technology, Intelligent Technology, Industrial Information Integration (pp. 95–99). IEEE. https://doi.org/10.1109/ICIICII.2017.47 *Yang, M., Shao, Z., Liu, Q., & Liu, C. (2017). Understanding the quality factors that influence the continuance intention of students toward participation in MOOCs. Educational Technology Research and Development, 65, 1195–1214. https://doi.org/10.1007/s11423-017-9513-6 Yu, A. Y., Tian, S. W., Vogel, D., & Kwok, R. C.-W. (2010). Can learning be virtually boosted? An investigation of online social networking impacts. Computers & Education, 55(4), 1494–1503. https://doi.org/10.1016/j.compedu.2010.06.015 Zayapragassarazan, Z. (2020). COVID-19: Strategies for online engagement of remote learners. F1000Research, 9. https://f1000research.com/documents/9-273 *Zhang, Y., Fang, Y., Wei, K. K., & Wang, Z. (2012). Promoting the intention of students to continue their participation in e-learning systems. Information Technology & People, 25(4), 356–375. https://doi.org/10.1108/09593841211278776 *Zhou, J. (2017). Exploring the factors affecting learners’ continuance intention of MOOCs for online collaborative learning: An extended ECM perspective. Australasian Journal of Educational Technology, 33(5). https://doi.org/10.14742/ajet.2914 Zhou, M. (2016). Chinese university students' acceptance of MOOCs: A self-determination perspective. Computers & Education, 92, 194–203. https://doi.org/10.1016/j.compedu.2015.10.012 *Zhu, Y., Zhang, J. H., Au, W., & Yates, G. (2020). University students’ online learning attitudes and continuous intention to undertake online courses: A self-regulated learning perspective. Educational Technology Research and Development, 68, 1485–1519. https://doi.org/10.1007/s11423-020-09753- w Corresponding author: Yan Zhang, zhy@ucas.ac.cn Copyright: Articles published in the Australasian Journal of Educational Technology (AJET) are available under Creative Commons Attribution Non-Commercial No Derivatives Licence (CC BY-NC-ND 4.0). Authors retain copyright in their work and grant AJET right of first publication under CC BY-NC-ND 4.0. Please cite as: Zhang, M., Li, S., & Zhang, Y. (2023). A meta-analysis of the moderating role of prior learning experience and mandatory participation on factors influencing MOOC learners’ continuance intention. Australasian Journal of Educational Technology, 39(2), 115-141. https://doi.org/10.14742/ajet.7795 Australasian Journal of Educational Technology, 2023, 39(1). 138 Appendix A: Overview of conclusions in sample studies Studies CON PU SAT PEOU SP PE OE SE ATT SI TTF Alraimi et al., 2015 √ × √ √ Brahmasrene & Lee, 2012 √ Chang et al., 2015(1) × √ Chang et al., 2015(2) × √ Chauhan et al., 2022 √ √ √ √ Chen et al., 2018 √ √ Cheng, 2019 √ √ √ √ Cheng, 2022 √ Chiu & Wang, 2008 √ √ √ √ × Dağhan & Akkoyunlu, 2016 √ √ √ Dai et al., 2020 √ √ √ Dai et al., 2022 √ √ √ Dai, Teo, Rappa, et al., 2020 √ √ √ Dalvi-Esfahani et al., 2020 √ √ √ × √ √ Daneji et al., 2019 √ × √ de Melo Pereira et al., 2015 √ Gu et al., 2021 √ √ √ Guo et al., 2016 √ √ √ Hsu et al., 2018 √ √ √ √ Ishak & Malaysia, 2020 √ √ √ Jo, 2018 √ √ √ Joo et al., 2018 √ √ √ Jung & Lee, 2018 √ × √ Kim & Song, 2022 √ × √ Lai & Lai, 2014 √ √ √ M. C. Lee, 2010 √ √ √ √ √ K. M. Lin, 2011 √ × × √ K. M. Lin et al., 2011 × √ √ √ W.-S. Lin & Wang, 2012 √ √ √ Lu et al., 2019 √ √ √ Luo et al., 2018 √ √ √ Najmul Islam, 2011 √ √ √ Nong et al., 2022 √ √ √ Nugroho et al., 2019 × √ Park et al., 2022 √ Qi et al., 2020 √ √ Ramayah & Lee, 2012 √ Rodríguez-Ardura & Meseguer- Artola, 2016 √ √ √ Rohan et al., 2021 √ √ √ Shanshan & Wenfei, 2022 √ √ Shao, 2018 √ √ Australasian Journal of Educational Technology, 2023, 39(1). 139 Suriazdin et al., 2022 √ × √ Tan & Shao, 2015 √ √ √ Tawafak et al., 2018 √ √ √ √ √ Tsai et al., 2018 √ L.-Y.-K. Wang et al., 2019 × × √ √ T. Wang et al., 2021 √ √ √ √ Wu & Chen, 2017 √ × √ √ Xu & Wang, 2017 √ √ √ √ Yang et al., 2017 √ √ Zhang et al., 2012 √ √ √ J. Zhou, 2017 √ √ √ √ Zhu et al., 2020 √ √ 20 29 31 12 5 7 7 7 11 4 8 × 0 6 1 5 2 1 0 0 0 1 0 Note. CON: confirmation; PU: perceived usefulness; SAT: satisfaction; PEOU: perceived ease of use; SP: social presence; PE: perceived enjoyment; OE: outcome expectation; SE: self-efficacy; SI: social influence; ATT: attitude. √: The paper studied this factor and found it has a significant effect. ×: The paper studied this factor and found it has an insignificant effect. Numerals (1) and (2) indicate two studies from the same article. Australasian Journal of Educational Technology, 2023, 39(1). 140 Appendix B: Overview of the articles in meta-analysis Studies Sample size Experience Mandatory participation Alraimi et al., 2015 316 Less Self-paced Brahmasrene and Lee, 2012 872 Less Self-paced Chang et al., 2015(1) 397 Less Self-paced Chang et al., 2015(2) 273 Less Self-paced Chauhan et al., 2022 396 Less Mandatory Chen et al., 2018 854 More Mandatory Cheng, 2019 391 More Mandatory Cheng, 2022 307 Less Mandatory Chiu & Wang, 2008 286 More Mandatory Dağhan & Akkoyunlu, 2016 467 Less Mandatory Dai et al., 2020 638 More Self-paced Dai et al., 2022 439 Less Self-paced Dai, Teo, Rappa, et al., 2020 306 More Self-paced Dalvi-Esfahani et al., 2020 456 More Mandatory Daneji et al., 2019 368 Less Mandatory de Melo Pereira et al., 2015 343 More Mandatory Gu et al., 2021 550 Less Self-paced Guo et al., 2016 244 More Mandatory Hsu et al., 2018 357 Less Self-paced Ishak & Malaysia, 2020 250 More Mandatory Jo, 2018 237 More Mandatory Joo et al., 2018 222 More Mandatory Jung & Lee, 2018 306 Less Mandatory Kim & Song, 2022 252 Less Mandatory Lai & Lai, 2014 240 Less Self-paced M. C. Lee, 2010 363 More Mandatory K. M. Lin, 2011 135 More Mandatory K. M. Lin et al., 2011 230 More Mandatory W.-S. Lin & Wang, 2012 88 More Mandatory Lu et al., 2019 300 More Self-paced Luo et al., 2018 258 More Mandatory Najmul Islam, 2011 175 Less Mandatory Nong et al., 2022 410 Less Self-paced Nugroho et al., 2019 48 Less Mandatory Park et al., 2022 224 More Self-paced Qi et al., 2020 372 Less Self-paced Ramayah & Lee, 2012 250 Less Mandatory Rodríguez-Ardura & Meseguer-Artola, 2016 2530 More Mandatory Rohan et al., 2021 206 More Self-paced Shanshan & Wenfei, 2022 555 Less Self-paced Shao, 2018 247 Less Self-paced Australasian Journal of Educational Technology, 2023, 39(1). 141 Suriazdin et al., 2022 164 Less Self-paced Tan & Shao, 2015 1347 Less Mandatory Tawafak et al., 2018 295 More Mandatory Tsai et al., 2018 126 Less Mandatory L.-Y.-K. Wang et al., 2019 170 More Mandatory T. Wang et al., 2021 854 Less Self-paced Wu & Chen, 2017 252 More Self-paced Xu & Wang, 2017 151 Less Self-paced Yang et al., 2017 294 More Self-paced Zhang et al., 2012 144 More Mandatory Zhou, 2017 435 More Self-paced Zhu et al., 2020 94 Less Mandatory Note. The numerals (1) and (2) indicate two studies from the same article.