Introduction

Online learning has been gaining in popularity evidenced by the growing number of online student enrolments and online education providers. This growing trend has been fuelled by the recent outbreak of the COVID-19 pandemic which forced educational institutions worldwide to move their classes online. Due to its capability to break down geographical, spatial, and temporal barriers, online learning affords the anytime, anywhere and for everyone philosophy of education and offers students a viable alternative to the traditional face-to-face learning mode (Kaufmann & Buckner, 2019; Naidu, 2017). Given the importance of lifelong learning that has been increasingly recognised by global governments, online learning is expected to be one of the fastest growing areas of education (Farrell & Brunton, 2020). A critical challenge remains as to how to increase student retention in online learning since the attrition rate in this learning environment is by far greater than that in the face-to-face learning mode (Hoi, 2022; Hoi & Hang, 2021a; Redmond et al., 2018). This high attrition rate is mainly attributable to students’ feelings of isolation and disconnection, and hence lack of motivation and engagement in a learning environment where students are expected to exercise greater self-regulatory skills and responsibility for their own learning than in the conventional brick-and-mortar classroom (Hartnett, 2015; Hsu et al., 2019; Lee et al., 2015; Redmond et al., 2018). In this context, understanding salient online learning environment factors and how they affect students’ motivation and engagement plays a crucial role in enhancing student retention and reducing drop-out rates in online learning.

To this end, Wang et al. (2020) proposed the development-in-sociocultural context model of student engagement as an integrative theoretical framework to delineate the engagement process of students in developmental contexts. This model is a comprehensive theoretical framework that synthesises research over the past 40 years on student engagement in various contexts and across various disciplines. It contributes to reducing the fragmentation associated with student engagement research by presenting a systematic and clear conceptualisation and clarification of the multidimensional student engagement construct. In this model, engagement is positioned in an ecological process during which various contextual and motivational factors interact to shape the learning process. Within this process, engagement plays multiple roles in the educational development of students and manifests itself to be a prime target for intervention programs that have potential to enhance students’ academic achievements and well-being and reduce students’ disengagement and drop-out. The model is graphically illustrated in Fig. 1 and further discussed in Sect. "Student engagement", "Key characteristics of student engagement", and "The development-in-sociocultural context model of student engagement".

Fig. 1
figure 1

The development-in-sociocultural context model. (Reproduced with permission from Wang et al. (2020)

Full size image

Drawing on this model, this study aims to make contribution to the online learning engagement literature by examining the effects of four online learning environment variables, namely teacher support, course clarity, student connectedness, and task relevance on student engagement in online learning, and what roles do students’ expectancy and task value beliefs play in these relationships. A detailed discussion of these variables and how their relationships are informed by the development-in-sociocultural context model of student engagement is presented in the next section.

Literature review

Student engagement

Student engagement is a critical psychological concept that emerged out of the drop-out prevention and school reform research traditions almost four decades ago (Wong & Liem, 2021). Much empirical evidence in this line of research suggests that engagement is a proximal predictor of school success, student achievement, and retention, thus manifesting itself to be a prime target for school interventions (Fredricks et al., 2004). Wang et al. (2020) define engagement as a student’s active involvement and interactions with learning activities, processes and contexts, and conceptualise it as a multidimensional construct consisting of behavioural, cognitive, and affective aspects (Fredricks et al., 2004; Wang et al., 2020). Behavioural engagement refers to the observable behaviours such as effort, concentration, persistence, attention, and hard work that students exhibit while participating in learning (Fredricks et al., 2004; Redmond et al., 2018). Cognitive engagement is concerned with students’ mental efforts and investment in learning including self-regulation, a desire to go beyond what is required, and cognitive strategies to understand complex concepts (Wang et al., 2016). Affective engagement is defined as students’ positive emotional reactions to learning activities, context, and their own participation in those activities (Fredricks et al., 2016). Although students might display different levels of each engagement component in a specific learning activity, engagement necessitates all three components, thereby painting a clear portrait of how students act, feel, and think during learning (Wang et al., 2020).

Key characteristics of student engagement

In addition to the multidimensional structure of engagement, the research literature highlights three other evidence-based characteristics of the engagement construct. First, there is ongoing debate as to whether engagement and disengagement are distinct and thus should be measured along a continuum with engagement constituting one end and disengagement constituting the other end or should be placed on two separate continua. This study follows the majority of recent studies on online learning environments by measuring engagement and disengagement along a single scale with higher values on the scale indicating engagement and lower values indicating disengagement (see, for example, Chiu, 2021a; Deng et al., 2020; Hoi & Hang, 2021a). Second, engagement operates across multilevel contexts and various timescales. The former refers to the sociocultural, family, school, classroom, peer, and task factors that shape student engagement while the latter is concerned with the moment-to-moment, day-to-day, or longitudinal timeframes in which engagement is measured. Since this study aims to identify online classroom variables that might have immediate effects on student engagement in the online learning classroom, the contextual variables of interest are those directly related to teachers, peers, and learning tasks that students interact with. Finally, engagement is a malleable concept, which is responsive to contextual and personal factors. It is this feature of engagement that attracts great research interest as researchers seek to understand factors that affect engagement and how to orchestrate those factors to enhance student engagement in the classroom.

The development-in-sociocultural context model of student engagement

As per the development-in-sociocultural context model (see Fig. 1), engagement plays multiple roles in the educational development of a student (Wang et al., 2019, 2020). More specifically, it is considered a pathway to student resilience, a facilitator of academic achievement and well-being, a mediating factor channelling the effects of contextual variables on students’ learning outcomes, and an academic outcome in its own right. In this study, engagement is examined as a desirable academic outcome, and thus the study focus is on how this outcome can be fostered in the online learning environment (i.e., the left portion of the model in Fig. 1). As an academic outcome, engagement is directly influenced by students’ developmental competencies and psychological beliefs. The former refers to students’ evolving social, emotional, and cognitive skills in their interactions with the social environments while the latter refers to their self-appraisals about learning and learning-related tasks that create a motivational context informing students about whether to remain persistent or give up in the face of setbacks. To understand this motivational context, Wang et al. (2020) integrated several salient motivational theories such as basic psychological needs theory (Ryan & Deci, 2017) (see Sect. "Online learning environment variables and student engagement") and expectancy-value theory (Eccles & Wigfield, 2002) (see Sect. "Expectancy-value beliefs as mediators") within the development-in-sociocultural context model. These motivational constructs are placed within the motivational beliefs and self-appraisals box that leads to the engagement box in Fig. 1. These psychological beliefs, in turn, are shaped by the students’ socialisation experience in class, school, family, or peer contexts. As this study is focused on the motivational mechanisms through which online learning environment variables influence student engagement, it explicitly draws on the basic psychological needs theory to identify salient online learning environment variables and then employs expectancy-value theory to hypothesise the motivational pathways through which these variables affect engagement.

Online learning environment variables and student engagement

Engagement does not exist in a vacuum, but instead emerges out of the interaction between the learners, the learning contexts, and learning outcomes (Skinner, 2016; Wang et al., 2020). Therefore, it is a highly situation-specific and context-dependent construct which is shaped by the objective environmental conditions and students’ subjective self-appraisals of those objective conditions. Identifying these contextual variables in the online learning environment and the motivational pathways through which they exert effects on student engagement plays a crucial role in enabling teachers to design and apply instructional strategies that have potential to foster student engagement in online learning. To this end, the development-in-sociocultural context model (Wang et al., 2020) posits that engagement flourishes in learning environments where students’ needs for competence, autonomy, and relatedness are fulfilled. Competence refers to one’s need to feel a sense of mastery and efficacy in one’s interaction with the social environment (Elliot & Dweck, 2005; Van den Broeck et al., 2016). Autonomy is concerned with students’ need to have a sense of ownership, volition, and self-endorsement of their own behaviours in learning rather than being controlled by external forces (Ryan & Deci, 2021). Relatedness is defined as one’s need to feel connected to others and a sense of belonging in one’s community (Ryan & Deci, 2000). Previous studies have identified various contextual variables that met students’ basic needs of competence, autonomy, and relatedness, thereby promoting student engagement in online learning environments (for a comprehensive review, see Chiu, 2021a; Hoi, 2022; Martin et al., 2018). Four such variables are the central foci of this study, namely teacher support, student connectedness, task relevance, and course clarity (i.e., these variables can be placed within the school context portion in the sociocultural characteristics box that leads to the motivational beliefs and self-appraisals box in Fig. 1). These variables are selected because they represent essential online learning classroom variables that directly influence students’ experience in the online classroom and are under the teachers’ control. By focusing specifically on these variables, the study has potential to generate pedagogical implications for the development and delivery of an engaging online learning classroom.

Course clarity refers to the extent to which the course organisation and instructional information are accessible and enable students to easily navigate and master the course content (Heilporn et al., 2021; Kaufmann et al., 2016). Course clarity fulfills students’ need for competence by giving them clear guidance about what it takes to become successful and achieve their desired learning outcomes in the online learning environment (Hoi, 2022). Studies have found that online learning courses which bolster students’ sense of competence by offering clear communication of expectations for student behaviours and clear guidance for learning positively affected all dimensions of student engagement (Chiu, 2021a; Martin et al., 2018).

Task relevance is related to the design and delivery of online learning tasks that match students’ personal interests and future goals beyond the online courses (Alamri et al., 2020; Lee et al., 2015). For example, students who intend to become a tour guide in the future are likely to have high perceived relevance of tasks that require them to make a presentation to introduce their classmates to historical and scenic spots in their hometown. Students are likely to feel a sense of autonomy and show greater participation in online learning when they are given tasks that are not only academically, professionally, and personally intriguing but also relevant to their future goals (Heilporn et al., 2021; Lee et al., 2015). Indeed, studies have found that students exerted greater effort, enthusiasm, and commitment to online learning when they were presented with relevant tasks and learning materials (Alamri et al., 2020; Chiu, 2021a; Martin et al., 2018).

Students’ need for relatedness is fulfilled in online learning environments where they perceive teachers to have supportive behaviours and feel connected to other peers. Teachers perceived to have supportive behaviours are those who show concern, care, respect, and understanding toward students and who are approachable (Kaufmann et al., 2016; Lan & Hew, 2020). Similarly, student connectedness refers to the extent to which students perceive their peers to be supportive, cooperative, and respectful (Kaufmann et al., 2016). Once students feel that they have a sense of personal contact and someone to go to when they have learning difficulties or personal problems, they are likely to form a positive relationship with teachers and peers and a sense of belonging in the online learning community (Stone & Springer, 2019). Teacher support and student connectedness were found in previous studies to positively predict students’ behavioural, cognitive, affective, and social engagement in online learning (Chiu, 2021a; Cole et al., 2019; Luan et al., 2020).

Expectancy-value beliefs as mediators

The development-in-sociocultural context model also proposes the expectancy-value theory (Eccles & Wigfield, 2002) as a motivational construct that informs our understanding of the student engagement process. According to the expectancy-value theory, the values that students attach to the learning activities and instructional practices (i.e., subjective task value) and their beliefs about the immediate or future success in such practices (i.e., expectancy belief) play a mediational role in the relationship between contextual conditions and students’ achievement-related outcomes (i.e., student engagement) (Eccles & Wigfield, 2020). While previous research provided strong evidence for the mediating role of basic psychological needs in the relationship between online learning environment conditions and student engagement, little is known about whether students’ perceived expectancy value beliefs also play the same mediating mechanism.

As one of the most prominent motivational frameworks, the expectancy value theory has been widely employed by educational psychologists to explain students’ achievement-related choices (e.g., engagement and persistence) and academic performance in both traditional and online learning environments (Eccles & Wigfield, 2020). Expectancy refers to students’ beliefs about how well they will do on an upcoming task while subjective task value refers to the utility, attainment, and interest values that students ascribe to a given task (Eccles & Wigfield, 2002, 2020). Utility value refers to the perceived usefulness of taking a certain course of action or participating in a certain activity (e.g., a student intending to become a tour guide might find a presentation task useful because it helps them improve their oral presentation skills and enrich their knowledge of their own hometown). Attainment value implies the importance of performing well on a given task (e.g., a student intending to become a tour guide might find a presentation task important because being able to deliver a presentation about a tourist attraction is part of their daily job as a tour guide). Finally, interest value is the enjoyment one gains from doing a task (e.g., a student intending to become a tour guide might find a presentation task interesting because introducing people to new and interesting places is what makes them want to become a tourist guide) (Gladstone et al., 2022). These beliefs operate in tandem to inform students’ decision-making as to whether to persist or give up on a task (Wang et al., 2020). Indeed, growing evidence in online learning environments has indicated that the more confidence students have in their ability to successfully complete an online task (i.e., expectancy belief), the more effort, energy, and enthusiasm they commit to completing the task (i.e., behavioural and affective engagement) (Fryer & Bovee, 2016; Jung & Lee, 2018; Pellas, 2014). Similarly, it was found that students employed more advanced cognitive strategies and engaged more deeply in the online learning process (i.e., cognitive engagement) if they perceived online learning tasks to be enjoyable, useful, and important (i.e., task value beliefs) for their future goals and interests (Joo et al., 2013; Lai, 2021; Manwaring et al., 2017; Zhang & Liu, 2019). For these reasons it is hypothesised that:

Hypothesis 1

Students’ expectancy belief positively and significantly influences their behavioural, cognitive, and affective engagement in online learning. (see Fig. 2)

Hypothesis 2

Students’ task value belief positively and significantly influences their behavioural, cognitive, and affective engagement in online learning

Fig. 2
figure 2

The hypothesised structural model

Full size image

Students’ expectancy-value beliefs are in turn shaped by their experience and interpretation of various social influences such as their past achievement-related events or other socialisers’ behaviours, attitudes, and expectations (Putwain et al., 2019; Trautwein et al., 2012). In the online classroom, factors such as course clarity, task relevance, teacher support, and student connectedness were all found to be proximal predictors of students’ expectancy and task value beliefs. For example, students were found to be more confident in their ability to succeed in online learning if the course is well-structured and offers clear guidelines for the navigation and mastery of course content (Martin et al., 2018; Trespalacios et al., 2023). On the other hand, students feel less competent and discouraged in online classes where they find the instructional videos unclear or the lesson content too difficult to understand (Chiu, 2021b). In terms of task relevance, Lan and Hew (2020) reported that students enjoyed a higher sense of competence and became more confident when they could see a clear connection between knowledge presented in the online course and how it could be applied in real-world contexts. Studies also indicated a positive relationship between student expectancy beliefs and the extent to which they perceived teachers and peers to be supportive and cooperative (Farrell & Brunton, 2020; Xu et al., 2020). Therefore, the following hypothesis is fomulated:

Hypothesis 3

Course clarity, task relevance, teacher support, and student connectedness positively and significantly influence students’ expectancy belief. (see Fig. 2)

Task value beliefs too are shaped by students’ interpretation of their own socialisation experience in the online classroom. Several studies have suggested that online learning activities that have immediate connections with real-world contexts or students’ personal interests and future career goals are considered by students to be enjoyable, useful, and important (Lee, 2015; Vanslambrouck et al., 2018). Similarly, studies have also highlighted the relationships between teacher support and student connectedness and the values that students attached to online learning tasks (Edwards, 2021; Farrell & Brunton, 2020; Lai, 2021; Mullen & Tallent-Runnels, 2006). In terms of course clarity, it is unknown whether course clarity influences students’ perception of the values of online learning. However, evidence from school-based research suggests that when students are provided clear guidelines for learning and consistent responses to their queries, they come to know what it means and what it takes to thrive in that context, thus attaching greater values to their learning (Assor et al., 2002; Wang & Eccles, 2013). It is believed that a similar proposition is also supported in online learning given students’ need to self-navigate their own learning journey in an environment where they have minimal face-to-face interaction with other social partners. For the reasons discussed above, it is hypothesised in the current study that:

Hypothesis 4

Course clarity, task relevance, teacher support, and student connectedness positively and significantly influence students’ task value belief. (see Fig. 2)

Hypothesis 5

Expectancy belief mediates the effects of teacher support, course clarity, student connectedness, and task relevance on students’ behavioural, cognitive, and affective engagement respectively.

Hypothesis 6

Task value belief mediates the effects of teacher support, course clarity, student connectedness, and task relevance on students’ behavioural, cognitive, and affective engagement respectively.

Methodology

Study context

The study was conducted in the context of various online undergraduate programs at a large open university in southern Vietnam. Students were from different academic disciplines including English language, business administration, finance and banking, law, and accounting. The online programs in which these students enrolled were offered over a 12-week period during which students were provided with learning and instructional materials, engaged in weekly discussion boards, peer review activities and exercises, and participated in three synchronous video conferences where they could discuss the course content with the instructors and their peers. Students’ course performance was assessed via three assignments which, along with their weekly activities, accounted for 35% of the course outcome while the final exam, either organised face-to-face or online, constituted the remaining 65%.

Participants

Data collection took place at the end of the second semester of the 2020–2021 academic year during which an online questionnaire was distributed to 517 students who gave their consent to participate in the study. Three hundred and fifty-one students completed and submitted the questionnaire yielding a response rate of 68%. Of these participants, 67.5% was female (N = 237) and 32.2% was male students (N = 113) with an age range of 19 to 65 (M = 32,64, SD = 8.03); first-year students constituted 43.9% (N = 154) of the sample while 26.5% (N = 93), 9.1% (N = 32), and 20.5% (N = 72) respectively were second, third-, and fourth-year students. All participants reported to have been fully equipped with computers and internet access and to have a minimum required level of computer literacy for online learning.

Instruments

Students’ perception of the online learning classroom (i.e., course clarity, task relevance, teacher support, and student connectedness) was measured by four scales adapted from previously validated instruments (Hoi, 2022; Kaufmann et al., 2016; Wang & Eccles, 2013). Course clarity was assessed by five items eliciting students’ perception of the extent to which the course provided clear guidelines for learning and instructions for the navigation and mastery of course content (a sample item is Based on my experience with and perception of this online course, I think the organisation of the course was clear). Three items were used to assess students’ perception of the relevance of the online learning tasks to their daily life and career goals (A sample item is How often do you learn things that are related to your life outside the online classroom). Teacher support was measured by six items about the degree to which students perceived their online teachers to be approachable and emotionally supportive (a sample item is Based on my online class interactions with the instructors, I perceived my instructor as respectful to me). Finally, students’ sense of connectedness with their peers in the online classroom was evaluated by four items (a sample item is Based on my online class interactions with the students in my class, I perceived students as cooperative with one another).

Students’ expectancy value beliefs in the online learning environment were measured by two scales adapted from Lazarides et al., 2020 and Midgley et al., 2000. The expectancy belief scale consists of five items assessing students’ perception of their ability to thrive in the online learning environment (a sample item is I’m certain I can master the skills taught in the online classes this year). The value belief scale includes four items measuring students’ perception of the utility (e.g., Learning the content of this online course is useful for what I want to do after I graduate and go to work.), importance (e.g., It is important for me to be someone who is good at the subject matter that I am learning in this online course), and interest (e.g., Learning the material covered in this online course is enjoyable.).

Student engagement in online learning was measured by a 12-item scale from Hoi and Hang (2021a, 2021b). Students’ observable behaviours in online learning was assessed by four items (e.g., I take notes when I participate in video conferences and discussion boards). Four items were used to gauge students’ cognitive investment and use of advanced strategies to understand online course content (e.g., I try to connect what I am learning in course materials with things I learn before). Finally, students’ affective responses to the online learning activities were indexed by four items (e.g., I enjoy online learning activities).

Data analysis

Structural equation modelling, a latent variable modelling technique, was employed to test the hypothesised model. A two-step modelling procedure was adopted: validation of a measurement model in which the relationship among the observed indicators and their underlying latent factor was scrutinised, followed by evaluation of a structural model in which the causal relationships among the latent factors were examined (Byrne, 2016). The former was handled through confirmatory factor analysis that allowed for the inspection of the model fit, item properties, convergent validity, and discriminant validity of the students’ perception of the online learning environment, expectancy value beliefs, and engagement scales. Model fit was evaluated based on well-established model fit indices suggested by Hu and Bentler (1999) and Byrne (2016): normed χ2 (< 3), Comparative Fit Index (CFI > 0.90), Tucker Lewis Index (TLI > 0.90), Root Mean Square Error of Approximation (RMSEA < 0.08), and Standardised Root Mean Square Residuals (SRMR < 0.07). Satisfactory item properties (i.e., item loadings, variance, and covariance) are characterised by statistically significant values and the absence of Heywood cases (i.e., negative variances or correlations higher than 1). Convergent validity, the extent to which the items converge to measure a latent construct, is achieved if the Average Variance Extracted (AVE) value for each construct is higher than 0.50 (Fornell & Larcker, 1981) while discriminant validity, the extent to which the latent constructs are distinguished from one another, is satisfied if the Heterotrait-Monotrait Ratio of Correlations (HTMT) is lower than 0.85 (Kline, 2016). To examine the reliability of the measures used in the study, composite reliability was computed for each measure. This reliability coefficient is computed based on the factor loadings of indicators obtained from confirmatory factor analysis, using the formula: CR = (sum of standardised loadings)2 / (sum of standardised loadings)2 + (sum of indicator measurement error) (Hair et al., 2014). Composite reliability values higher than 0.70 are considered acceptable (Hair et al., 2014).

Evaluation of the structural model followed the same model fit testing procedure as the measurement model. Once this fit is established, the structural paths among the latent factors could be examined. Standardised values were used to determine the relative strength of the direct relationship among the variables while the indirect effects of online learning environment variables on student engagement were assessed via the bias-corrected bootstrapped 95% confidence intervals (CIs) based on 5000 random samples (Preacher et al., 2007). Bootstrapped CIs that do not cross zero suggest significant indirect effects. All analyses were conducted in AMOS 28 using the maximum likelihood estimation method. To enable the comparison among the path coefficients in the structural model and to provide the context for interpreting the magnitude of the effects on learning outcomes in educational research, we follow Keith (2019) in interpreting effect sizes as too small to be meaningful (β < 0.05), small but meaningful (0.05 < β < 0.10), moderate (0.10 < β < 0.25), and large (β > 0.25) (Keith, 2019, p. 63).

Results

Table 1 presents descriptive statistics and parameter estimates of the measurement model. The skewness and kurtosis values were all within the acceptable range of |3| and |7|, suggesting the univariate normal distribution of data. The measurement model in which all latent constructs were allowed to covary with each other showed adequate fit (χ2/DF = 1.879; CFI = 0.934; TLI = 0.927; RMSEA = 0.050, CI [0.046—0.054]; SRMR = 0.049).

Table 1 Parameter estimates of the measurement model
Full size table

All standardised parameters were statistically significant, and no Heywood cases were identified, verifying the psychometric qualities of the scale items. The convergent validity of the scales was supported by strong loadings (0.488—0.891) of each item set onto their respective factor and acceptable AVE values (0.544—0.712) for all scales (except for the behavioural engagement scale but see below). The lower-than-optimum AVE value of the behavioural engagement scale was of little concern because the composite reliability measure of this scale (0.784) was higher than 0.60 (Fornell & Larcker, 1981). The discriminant validity of the scales was also supported given that all HTMT correlations were lower than 0.85 (0.217—0.699) (see Table 2).

Table 2 HTMT correlations among the constructs
Full size table

The structural model (see Fig. 3) achieved an acceptable level of fit to data (χ2/DF = 1.996; CFI = 0.923; TLI = 0.917; RMSEA = 0.053, CI [0.049—0.057]; SRMR = 0.064). Subsequent examination of the hypothesised relationships among the study variables revealed that students’ expectancy and subjective task value beliefs positively predicted all dimensions of student engagement. Expectancy belief was a more robust predictor of students’ behavioural, cognitive, and affective engagement than value belief (βexpt = 0.40, 0.45, 0.55 respectively; p < 0.001, large effect size; βvalue = 0.22, 0.12, 0.26 respectively; p < 0.05, moderate effect size). Of the four online learning classroom variables, only course clarity and task relevance had positive direct effects on both expectancy (βcourse = 0.248, p < 0.001; βtask = 0.317, p < 0.001, moderate effect size) and value beliefs (βcourse = 0.209, p < 0.01; βtask = 0.349, p < 0.001, moderate effect size). Teacher support only had positive effect on expectancy belief (βteacher = 0.209, p < 0.01, moderate effect size) while student connectedness was found to predict neither expectancy nor value beliefs.

Fig. 3
figure 3

The structural model with standardised estimates

Full size image

The indirect effects of course clarity (95% CIs = [0.072—0.264], β = 0.144; [0.063—0.262], β = 0.137; [0.096—0.345], β = 0.191, moderate effect sizes) and task relevance (95% CIs = [0.112—292], β = 0.202, moderate effect size; [0.108—0.277], β = 0.186, moderate effect size; [0.156—0.381], β = 0.265, large effect size) respectively on students’ behavioural, cognitive, and affective engagement via expectancy and task value beliefs were all significant. On the other hand, teacher support indirectly affected student engagement only via expectancy belief (95% CIs = [0.025—0.170], β = 0.097, small but meaningful effect size; [0.024—0.149], β = 0.109, moderate effect size; [0.026—0.201], β = 0.134, moderate effect size) respectively on behaviour, cognitive, and affective). Finally, both students’ expectancy and task value beliefs were found to have no mediating effect on the relationship between student connectedness and student engagement.

Discussion

This study was situated in an online learning context characterised in the literature by the isolated and disconnected feelings that students frequently experience and the high dropout rate that online education providers are trying to reduce (Hoi & Hang, 2021a). Its purpose was to unpack salient online learning classroom variables that might influence student engagement and the motivational mechanisms through which these influences are exerted, thereby suggesting methods to leverage student engagement and retention in online learning. Table 3 summarises the research findings in relation to the research hypotheses.

Table 3 Summary of the research findings in relation to the research hypotheses
Full size table

The findings largely supported the integration of the basic psychological need theory and expectancy value theory into an overarching framework of engagement that has potential to inform research on student engagement. Students’ expectancy and task value beliefs were found to be significant predictors of student engagement in online learning and in turn were significantly predicted by online learning classroom variables that fulfill students’ needs for competence, autonomy, and relatedness. However, contrary to the general literature on the expectancy value theory that attested to the role of expectancy belief as a more robust determinant of academic achievements (e.g., test results, GPA) and the role of task value belief as a stronger predictor of achievement-related choices (e.g., engagement and persistence) (Marsh et al., 2005; Wigfield & Eccles, 2000), the current study found that students’ expectancy belief predicted student engagement in online learning with larger magnitude than task value belief. While this finding represented a departure from the general literature, it might be explained by the contextual conditions that shaped expectancy and task value beliefs. In this respect, it was found in this study that students’ expectancy belief was significantly predicted by course clarity, task relevance, and teacher support while task value belief was influenced only by course clarity and task relevance. The differential effects of expectancy and task value beliefs on student engagement might also be explained by variables that had not been included in the current study. For example, in a recent study, Lai (2021) reported that the effect of task value on student engagement was weakened if students perceived learning tasks to be difficult – a variable that can be offset by high expectancy belief. Had perceived task difficulty been included as a contextual variable in this study, the results might have been different.

Task relevance was found to be the most robust determinant of both students’ expectancy value beliefs and engagement. This finding not only reenforced the positive effect of providing relatable and meaningful tasks on student engagement in online learning as reported in previous studies (see, for example, Chiu, 2021a; Lee et al., 2015; Martin et al., 2018; Vanslambrouck et al., 2018) but also extended on those studies by shedding light on the motivational mechanisms via which this effect was exerted. As such, students are likely to attach greater values to and become more confident in tasks that are directly relevant to their life and future career, which in turn promotes their engagement in online learning (Lee, 2015). Therefore, the design and delivery of tasks in the online learning environment should take this feature into account and should be exercised with caution because variables such as task difficulty might nullify the values of task relevance (Lai, 2021).

Similar to task relevance, course clarity indirectly influenced all dimensions of student engagement with moderate to large effect sizes via both expectancy and task value beliefs. This construct can be likened to the construct of structure, a dimension of teacher support identified in self-determination theory for classroom practice, which involves clear communication of expectations for student success, well-structured lessons, strong guidance during learning, and effective learning materials (Lietaert et al., 2015). While structure emphasises the role of teachers in delivering these types of support in the classroom, course clarity involves the organisation of course information in the online classroom that can be accessed by students at the beginning of the course. Just as structure was reported in previous classroom-based studies to promote different dimensions of student engagement (Skinner & Belmont, 1993; Wang & Eccles, 2013), it was found in this study that the extent to which the course organisation provided students with clear instructions on the use of learning materials and technology, and clear guidelines for assignments, positively predicted their engagement in online learning through the consolidation of their expectancy and task value belief. Therefore, an important element to consider when designing online courses is the clear communication of the course objectives as well as guidelines for learning and assessment.

Contrary to task relevance and course clarity, teacher support and student connectedness were found to have no or only partial effect (with small effect sizes) on student engagement via expectancy and task value beliefs. While this finding diverged from those consistently reported in conventional face-to-face classroom research that highlighted the positive effects of teacher support and student connectedness on student engagement (Sulis & Philp, 2020; Wang & Eccles, 2013), it received support from several studies in online learning environments (Chiu, 2021b; Edwards, 2021; Lan & Hew, 2020). Consistent with these studies, the finding suggests that students’ sense of connectedness with peers was largely undermined in the online learning context. Although the online courses were designed in a way that required students to collaborate for task completion, it could be inferred from the course information that this collaboration was largely asynchronous. The limited synchronous interactions, therefore, might have deprived students of opportunities to get regular support and engage in constant interaction with peers. In a context where students need to exercise more regulation and self-navigate their own learning journey through the entire course, lack of constant peer support and interaction when learning difficulties occur is likely to dampen their sense of connectedness and their subsequent engagement (Kaufmann & Vallade, 2020). Teacher support, however, was found to influence student engagement via expectancy belief rather than task value beliefs. A plausible explanation could be that the type of teacher support, as perceived by students, was of a reactive nature rather than proactive. That is, online teachers, due to limited interactions with students and to the nature of the course structure, might have provided support only when learning problems occur or when students approach them for advice or feedback during their self-regulated learning. This type of support might enable students to solve problems, give them confirmation as to whether they are on the right track, or give them more confidence in their self-regulated learning journey (i.e., expectancy belief). However, this type of support is not likely to arouse their interest, enjoyment, or perceived importance and usefulness of the tasks at hand (i.e., task value belief) compared to proactive support wherein teachers are actively involved in students’ work and offer support on a regular basis. While teacher support has been well researched in the literature on student engagement, the results of this study suggest that more research needs to be done to gain a richer understanding of the right type of teacher support that leverages student engagement in a specific context such as online learning.

Implications

The study findings offer important suggestions for the development of online courses that have potential to foster student engagement. First, since task relevance manifested itself to be the most salient determinant of student engagement in online learning, course designers and teachers should take students’ learning preferences, learning habits, personal interests, current professional practices, and future career orientation into account in an attempt to provide them with authentic and engaging tasks. Tasks that align course content and skills with students’ professional practices or daily life and that stimulate a genuine need among students to use skills and knowledge from the online course for task completion might be particularly engaging. Teachers making videos in which successful entrepreneurs or alumni are invited to share their start-up experience and then showing these videos in a business course, or projects that require students to make a video interview with native speakers of a foreign language in a language course are some examples of task relevance. In addition, since student connectedness was undermined in the online learning environment, the interactive nature of course content and activities should be improved. Some strategies that have been successfully employed in previous studies to enhance students’ sense of connectedness include teachers setting aside regular hours per week to answer students’ questions and provide feedback to common issues raised by students synchronously (Lan & Hew, 2020), or setting up and managing informal online space through social networks parallel to the online learning management system where they can make themselves more available to students (Hoi & Hang, 2021b). Tasks that necessitate interactions among students need to be manipulated in ways that students need to genuinely collaborate with each other synchronously for fulfillment rather than just to take turns and complete their parts. Encouraging students to contribute to discussion forums by building on others’ posts and rewarding them with plus points when their contribution is acknowledged by other classmates might also help improve students’ sense of competence in the online community.

Conclusions

This study sheds light on some salient online learning classroom variables that hold strong potential for fostering student engagement. It also contributes to theorising and conceptualising online learning engagement from an ecological perspective. Among the online learning environment variables, course clarity and task relevance play critical roles in motivating students to engage in online learning. On the other hand, online learning designers and teachers should put more effort into innovating the online course design, instructional strategies, and support features in ways that instigate stronger task value attachment and perceived competency among students, which in turn will contribute to more effective participation, collaboration, and involvement in online learning.

Limitations

Despite the positive findings, there are several limitations in this study that could be addressed in future research. First, the cross-sectional nature of data collection in this study did not allow for the examination of fluctuation of student engagement over the entire semester or academic year and how students develop their developmental competencies and psychological beliefs in response to the evolving online learning environment to remain engaged in learning and persistent in the face of setbacks. Future studies might need to investigate this longitudinal aspect of student engagement in online learning environments. Second, from an ecological perspective, student engagement constitutes a central role in the bidirectional or reciprocal relationships among the learning context, the self, and relevant learning outcomes. Therefore, hypothesising and testing the causal relationship among the online learning classroom variables, motivational beliefs, and engagement as done in the current study might have obscured the reciprocal effect of engagement on future motivational beliefs and learning environment. As suggested by Wang et al. (2020), engagement begets future engagement by bolstering students’ psychological beliefs and reinforcing students’ positive experience of the learning environment. This reciprocal relationship deserves further research into the online learning environment in future studies. Third, the study only focused on the online learning classroom variables, expectancy and task value beliefs, and online engagement. There are other contextual (e.g., sociocultural and family factors), developmental (e.g., social, emotional, and cognitive skills), and motivational (e.g., mindset) variables suggested in the development-in-sociocultural context model that are beyond the scope of the current study. Future studies are encouraged to examine these variables to paint a more comprehensive picture of the student engagement process. Finally, the study did not examine the effects of demographic variables such as gender, grades, and academic majors due to the unequal distribution of these variables in the convenience sample. Future studies might take these factors into account to afford a richer understanding of the individual differences that might affect the engagement process in online learning.