Perceived enjoyment is defined as the extent to which the activity of using a specific system is perceived to be enjoyable in its own right, aside from any performance consequences resulting from the use of the system (Venkatesh, 2000). Lee (2010) argues that e-learning often has many entertainment interactive functions, and that users can derive great enjoyment when using such systems. Given the findings in the literature cited above, it is clear that perceived enjoyment is a construct that needs further attention. In particular, we believe that insights about specific factors influencing students’ perceived enjoyment of online learning are necessary to enhance current approaches to teaching in the post-COVID-19 setting. Darby and Lang (2019) assert that students’ enjoyment of online learning is important, as online courses are proving to have higher attrition rates, and therefore when students enjoy their courses, they are more likely to continue engaging, and they will thus be retained in the online class. Recent literature stresses that understanding factors that influence students’ enjoyment of online learning is fundamental to promote student retention and success during and after the COVID-19 pandemic (Okada and Sheehy, 2020). Against this background, we investigate students’ perceived enjoyment and operationalize it as a dependent variable.

Worldwide, higher education institutions are realizing the importance of quality e-learning systems (Mulhem, 2020). Unfortunately, quality in the e-learning context seems complex, and, as such, the construct of perceived system quality has been operationalized in different ways by different authors (Alla, 2013). According to Dreheeb et al. (2015, p. 13), system quality refers to “the extent to which an industry defines a set of desirable features that should be incorporated into the product to enhance its lifetime performance.” DeLone and McLean (2003) postulate the following dimensions of system quality: adaptability, availability, reliability, response time and usability. Other dimensions are convenience, access, flexibility, integration, sophistication, system speed, navigation and network speed, and user interface consistency (Alla, 2013). Despite the fact that scholars have identified different dimensions of system quality, it is postulated that system quality affects both the user and user satisfaction (Mulhem, 2020), beliefs about the system, and the user’s intention to use the system (Ramayah et al., 2010). Hence, the hypothesis is that perceived system quality influences users’ perceived enjoyment of online learning.

Perceived usefulness refers to the level at which a technology’s end user believes that the technological tools used in teaching and learning will enhance the end user’s work and study performance in comparison to other methods of teaching and learning (Liu et al., 2009). According to Rizun and Strzelecki (2020), perceived usefulness influences the end user’s decision to accept or reject a particular technology. Enjoyment of something, or taking pleasure in it (e.g., online learning), is related to perceived usefulness (Cicha et al., 2021), and it positively affects learning (Hernik and Jaworska, 2018). Cicha et al. (2021) report that students who participated in their study believed that it was possible to gain more knowledge when learning online because of their knowledge of technological tools. Paradoxically though, their study revealed that the students were dissatisfied with online learning and did not enjoy it. Similarly, Rizun and Strzelecki (2020) found that students perceived online learning as useful due to its use of technological tools, but at the same time they reported enjoying and preferring studying in traditional educational settings, where they can interact with peers and academics, receive in-person feedback, and experience the campus environment. By contrast, a study by Pham et al. (2021) revealed that students perceived online learning during COVID-19 as useful, because it is exciting and time-saving and makes learning easier and less costly, and the students therefore enjoyed online learning. Thus, it can be concluded that students may not always find online learning through the use of technologies enjoyable, but may still perceive the use of technologies as useful (Camilleri and Camilleri, 2019). However, some studies have indicated a relationship between enjoyment and usefulness, and, as such, this relationship will be confirmed in a different setting, as indicated in the conceptual model.

Perceived ease of use refers to the level at which an end user believes a particular technology can be effortlessly and conveniently used (Esteban-Millat et al., 2018). According to Pham et al. (2021), online learning is designed to easily share and obtain knowledge on a day-to-day basis, and perceived ease of use in this context includes aspects such as accessibility, transmission speed, and learning adaptability. A study conducted during the COVID-19 outbreak found that students’ positive attitude toward acceptance and use of online learning was partly due to its perceived ease of use (Drueke et al., 2021). Drueke et al. (2021) found that factors of perceived ease of use, such as data availability, general media affinity, and technical infrastructure, improved the students’ positive attitude to accept online learning. Another study conducted during COVID-19, which included a sample of students who experienced a shift to completing online learning, revealed that there was an interdependent relationship between perceived ease of use and perceived enjoyment (Muñoz-Carril et al., 2021). The study found that perceived ease of use positively affected the students’ attitudes, causing perceived enjoyment of online learning. Similarly, Cicha et al. (2021) found that enjoyment correlated with ease of use of online teaching and learning technologies. Therefore, perceived ease of use is expected to be related to perceived enjoyment in the South African context.

Perceived cognitive benefits refer to improved understanding and application and positive perception of the learning material as a result of the online learning environment (Lee et al., 2010). Baturay (2011) found a strong positive relationship between cognitive learning reported and student satisfaction with an online course. The scholar further reports that cognitive learning and final exam scores were only moderately related, which indicates that good grades do not necessarily mean that cognitive learning has taken place at a deep level. For example, a student could memorize and understand learning content but may not be able to apply and analyze the content. However, due to the relationship established between cognitive learning and student satisfaction, it is postulated that perceived cognitive benefits will be positively related to perceived enjoyment.

Helyer (2015) states that reflection is about looking back at events that happened in the past, to obtain a better understanding of what happened, make sense of the experiences, and discover possible mistakes, to avoid repetition of the mistakes. Perceived reflective learning is also regarded as a process that leads to reflection on knowledge sources and experiences of students and collective groups, which may contribute to a better understanding of their learning journey (Colomer et al., 2013). Bubnys and Žydžiūnaitė (2010) suggest that reflective learning is a continual process that enhances learning outcomes for individuals and collectives. Chang (2019) supports the latter view and asserts that reflection in learning is a necessary process, as it provides an opportunity for students to revisit their learning, document the learning journey, and provide suggestions to improve the learning experiences of future students. Previous research (Invernizzi et al., 2019) has shown that perceived reflective learning positively contributes to enjoyment and satisfaction in face-to-face learning. It is suggested that perceived reflective learning will also lead to increased enjoyment of learning in online approaches.

Against this background, the following hypotheses are tested:

The aim of the study was to establish the extent to which selected factors related to the online learning experience influence perceived enjoyment of online teaching and learning. The reason for conducting the study was that higher education institutions were not prepared to transition to online teaching and learning, and this lack of preparedness could have influenced students’ enjoyment of the online learning experience. Furthermore, previous studies found that students perceived online teaching and learning negatively (Van der Walt and Nkoyi, 2022) which could contribute to a lack of student engagement, and high dropout rates (Baturay, 2011). Another reason is that there is a dearth of research studies focusing on the antecedents of perceived enjoyment of online teaching and learning. Therefore, the primary objective of the study was to establish whether selected variables, namely perceived system quality, perceived usefulness, perceived ease of use, perceived cognitive benefits and perceived reflective learning, influence perceived enjoyment of the online learning experience.

This study followed a deductive approach, because hypotheses based on previously conceived theories were tested (Saunders et al., 2012). It was deemed appropriate to use a quantitative research design, since the study aimed to describe the influence of selected factors related to online teaching and learning on perceived enjoyment thereof. To collect the quantitative data, an online survey was administered once-off using a structured questionnaire. Therefore, a cross-sectional design was utilized (Creswell and Creswell, 2018).

The target population of the study was students enrolled at traditional universities, universities of technology and comprehensive universities in South Africa. Convenience sampling was used, as recommended by Jordan et al. (2015), since only readily accessible students were requested to participate in the research study. The final sample consisted of 501 students.

The demographic information of the respondents is summarized in Table 1.

The sample comprised more females (62.5%) than males (36.7%). The main fields of study were management (16%), followed by IT (14.2%), education (13.4%), humanities (11.4%), engineering (11.2%), and law (10.0%). The majority of the respondents (86.4%) were first-year students (20.6%), second-year students (32.9%), and third-year students (32.9%). A total of 83% of the respondents were studying at a traditional university (39.9%) or a university of technology (43.1%).

The QuestionPro platform was used to create the online survey. A customized template was used on this platform to create the structured questionnaire which was used to collect the data. In the first section of the questionnaire, biographical questions were posed, which were used to describe the sample. Perceived enjoyment was measured using five items, an example being “Using online teaching and learning is pleasant.” Four items measured perceived usefulness, an example being “Online teaching and learning is useful in supporting my learning,” and three items measured perceived ease of use, an example being “Overall, I think the online program used by the university is easy to use.” Perceived cognitive benefits were measured with four items, an example being “Online teaching and learning helps me to understand the content easier,” and perceived reflective learning was measured with four items, an example being “Online teaching and learning enables me to reflect on my own understanding.” Perceived system quality was measured with five items, an example being “The information on my university’s e-learning website is up to date.” Responses were indicated on a five-point Likert scale, with response options ranging from “Strongly Disagree” to “Strongly Agree.” To gather data from a representative sample, the survey was distributed explicitly to students studying at a higher education institution in South Africa through selected social media channels and online platforms commonly used by these students. The survey was distributed to the sample in November 2020, and it remained open for 3 weeks.

SmartPLS version 3.2.9 was used to analyze the data. Both descriptive and inferential data analyses were done. To analyze the structural relationship between the variables included in the study, Structural Equation Modeling (SEM) was employed which combines factor analysis and multiple regressions (Ramlall, 2017). Data analysis comprised two stages: assessment of the measurement model (the inner model), followed by assessment of the structural model (the outer model). Hair et al. (2019) provide the following guidelines for assessing a reflective measurement model, which was the model followed in this study. First, item reliability must be assessed. The outer loading of items should be above 0.708 to show that the construct explains more than 50% of the variance in the item. Hair et al. (2011) present an exception to this guideline. They suggest that items with outer loadings less than 0.7 but higher than 0.4 can be retained in the measurement model as long as they do not threaten the reliability and validity of the measurement model. Next, Hair et al. (2019) state that the internal consistency reliability must be confirmed by inspecting the composite reliability (CR) and Cronbach’s alpha indices of each factor’s measures. For both indices, there was sufficient evidence of internal consistency reliability, as the outer loadings were higher than 0.7. The third step is to confirm the convergent validity of each construct’s measures. Evidence of convergent validity is that the average variance extracted (AVE) of the items measuring a construct is higher than 0.5. Lastly, the discriminant validity of the construct measures must be confirmed. Adequate discriminant validity between the construct measurements is when the squared inter-construct correlation (the shared variance between the two constructs) is not higher than the AVE of each of the two constructs.

Assessment of the structural model starts with inspecting the variance inflation factor (VIF) between the independent factors (Hair et al., 2019). The VIF should be close to 3.0 or lower. If collinearity is not an issue in the structural model, the R² of each endogenous factor can be examined to determine the in-sample explanatory power of the structural model. The guideline for interpreting the R²-value is that values of 0.75, 0.50, and 0.25 can be considered substantial, moderate and weak in-sample explanatory power, respectively (Hair et al., 2019). To accept or reject the hypotheses represented by structural paths in the structural model, the path coefficients and their statistical significance (α = 0.05, two-tailed) were examined. In partial least squares structural equation modeling, the p-values are obtained using the bootstrapping procedure (Hair et al., 2019). In this study, 5,000 subsamples were used in this procedure.

Ethical considerations such as informed consent, voluntary participation, confidentiality and anonymity were ensured by means of a participant information sheet, which was included in the research instrument. Questionnaires were completed anonymously to ensure confidentiality, and participants were assured that data would only be used for research purposes. Only the lead researcher had access to the questionnaire responses, which are password-protected.

The initial assessment of the measurement model yielded acceptable item reliability, internal consistency reliability and convergent validity. However, evidence of discriminant validity between the measurement of perceived cognitive benefits and perceived reflective learning was not supported by the results of the assessment of the measurement model. The cross-loading matrix was inspected to identify items that could contribute to the lack of discriminant validity between the two constructs. The results in the cross-loading matrix showed that the highest cross-loading was REFT3 on perceived cognitive benefits (0.772). The item was removed, and the results of the revised measurement model are reported in Table 2.

Table 2 summarizes the measurement model assessment results required to complete the first three steps in assessing the model. All outer loadings were statistically significant, but two outer loadings were less than 0.708 (REFT1 = 0.669; SYS1 = 0.567). Although these two items did not meet the criterion of item reliability, the items were retained in the measurement model, as they did not threaten the reliability and the validity of the model. As reported in Table 2, all constructs in the measurement model have a Cronbach’s alpha and a CR above 0.7, except for perceived reflective learning (Cronbach’s alpha = 0.670). The Cronbach’s alpha was deemed acceptable, as it was close to 0.7, taking into account that the Cronbach’s alpha is a not-so-precise measure and a conservative estimate of reliability (Hair et al., 2019), and the fact that the CR value exceeded 0.7. Also, the AVE of each construct was higher than 0.5. The measurement model assessment results in Table 2 provide sufficient evidence of internal consistency reliability and convergent validity.

Discriminant validity results are reported in Table 3.

Table 3 summarizes the AVE of constructs compared to the squared correlation between constructs, to assess the discriminant validity in the measurement model. As shown in Table 3, after excluding REFT3 from the measurement model, the AVE of the two constructs is higher for each pair of constructs than the squared correlation between the constructs. Thus, the revised measurement model showed acceptable discriminant validity to continue assessing the structural model.

To describe the student sample’s responses to the variables included in the study, measures of central tendency (mean, median, and standard deviation) were determined.

From the results presented in Table 4, it is evident that respondents perceived all the constructs measured negatively (i.e., comparing the mean score with the median or middle score). The respondents’ negative perception of perceived usefulness and perceived ease of use, show that they perceive online teaching and learning as problematic, possibly negatively affecting their engagement and performance. Their negative perception of cognitive benefits implies that they perceived the learning material, and the understanding and application thereof negatively. In terms of reflective thinking, their negative responses are indicative that they did not find online learning activities mentally challenging. The respondents’ negative assessment of system quality shows that they mostly disagreed that the system used by the respective university, is accessible, reliable, and available. Lastly, the results show that the respondents did not enjoy online teaching and learning.

All VIFs were less than 3.0. The highest VIF was 2.830. Thus, collinearity was not an issue that could invalidate the results of the assessment of the structural model.

The structural model assessment results are depicted in Figure 1.

The five determinants of perceived enjoyment explained 73.8% of the variance in perceived enjoyment. Thus, the in-sample accuracy of the structural model is substantial. The structural model assessment results in Figure 1 show that H1 was not accepted, as the influence of perceived system quality on perceived enjoyment was not statistically significant (p > 0.05). Both perceived usefulness and perceived ease of use influenced perceived enjoyment positively and statistically significantly (p < 0.05). Thus, H2 and H3 were accepted. The statistically significant positive influence of perceived usefulness on perceived enjoyment was stronger than the influence of perceived ease of use on perceived enjoyment. Overall, perceived cognitive benefits had the strongest positive statistically significant influence on perceived enjoyment. Based on the statistical significance of the positive influence, H4 was accepted. H5 was rejected, as the influence of perceived reflective learning on perceived enjoyment was not statistically significant (p > 0.05).

The findings of the study show that students’ perceived enjoyment of online teaching and learning is influenced by several factors. Most importantly, lecturers should ensure that course content is cognitively stimulating, and that learning management systems are used that are easy to operate and accessible to students despite data and connectivity challenges. Due to increased uncertainty in the external environment, the looming Fourth Industrial Revolution, and the fact that the student population is becoming more diverse, it is unlikely that higher education institutions will return to traditional face-to-face teaching and learning (Van der Walt and Nkoyi, 2022). As such, research must continue regarding students’ e-learning experience, their satisfaction and enjoyment thereof, and the instructional design of online courses to promote high-quality teaching and learning. Consideration should also be given to the perceived usefulness of online teaching and learning and perceived system quality. Very few studies have focused on variables that could possibly influence perceived enjoyment, and, as such, the current study contributes to existing knowledge regarding the intention of continued use of online teaching and learning.

Despite the noteworthy findings of this study, the study had some limitations. Although the intention was to draw a sample using simple random sampling, this was not possible, and we had to rely on convenience sampling. Since, the sample do not necessarily represent the population from which the sample was drawn, the findings cannot be generalized to a wider population. However, the sample was diverse in terms of the socio-demographic variables measured, and care was taken to confirm the measurement model. It is therefore recommended that future research should consider using simple random sampling or another more robust sampling method to avoid sampling and response bias. A cross-sectional research design was used and, as such, there is limited causality inference. However, the sample was adequate in size. It is further recommended that the established relationships be interrogated using qualitative research, to understand how the variables influence perceived enjoyment of online teaching and learning. More research is also needed to understand how perceived reflective learning can be used to enhance perceived enjoyment on an online platform.