Introduction

During the COVID-19 pandemic, traditional on-campus teaching and learning at conventional universities was severely limited and temporarily even impossible due to various preventive measures imposed by governments to diminish the spread of the coronavirus (Marinoni et al., 2020; Universities of The Netherlands, 2022). To ensure the continuation of education, universities rapidly implemented emergency remote teaching (ERT) with online or distance education becoming the norm (Aristovnik et al., 2020; de Boer, 2021; Mseleku, 2020). While few studies found no or even a positive effect of ERT on students’ academic performance (Gonzalez et al., 2020; Iglesias-Pradas et al., 2021; Talsma et al., 2021), most empirical findings indicate that ERT had a detrimental effect on students’ perceived academic achievement, grades, course completion, test scores, and study delay (Aucejo et al., 2020; Bird et al., 2022; Farcnik et al., 2021; Inspectorate of Education, 2021; Interstedelijk Studenten Overleg, 2020; Jiang, 2021; Kim et al., 2021; Mahdy, 2020; Oducado & Estoque, 2021; Tinjić & Halilić, 2020; van den Broek et al., 2022; Varachotisate et al., 2023). For example, the percentage of students experiencing study delay in the Netherlands was up to 8.8% higher after the onset of the pandemic compared to 2018 (Inspectorate of Education, 2021; Interstedelijk Studenten Overleg, 2020). However, little is known about the mechanisms underlying the relationship between ERT and impaired academic performance. Understanding the underlying mechanisms is important to develop strategies diminishing the detrimental effects of ERT on students’ academic performance.

Drawing on theoretical models that describe how study circumstances impact students’ academic performance, the performance-impairing effect of ERT might be explained by students’ study-related experiences, in particular students’ satisfaction and academic wellbeing (Duque, 2014; Lesener et al., 2020; Salmela-Aro et al., 2022a; Schertzer & Schertzer, 2004). Students’ (education) satisfaction refers to an attitude resulting from the evaluation of students’ educational experience (Weerasinghe et al., 2017). Students’ academic wellbeing is often studied by two core concepts, i.e. academic burnout and academic engagement (Lesener et al., 2020; Salmela-Aro et al., 2022a). Academic burnout is primarily characterised by emotional exhaustion due to study demands, accompanied by a cynical attitude towards one’s study and perceived academic inefficacy, while academic engagement is typified by high levels of energy, enthusiasm, and absorption (Schaufeli et al., 2002).

In the theoretical models proposed by Duque (2014) and Schertzer & Schertzer (2004), students’ satisfaction is considered to mediate the effect of the learning environment, particularly the match between students’ preferences and learning circumstances, on student’s academic performance; with a better match leading to more satisfaction, which in turn promotes academic performance. Several studies conducted during the pandemic found that (a higher amount of) ERT, which does not match conventional university students’ preferences, is related to lower education satisfaction (Lee et al., 2021; Loton et al., 2021; Vollmann et al., 2022; Wissing et al., 2022) and that lower satisfaction with ERT is associated with lower academic performance in terms of dropout intentions, grades, and perceived academic achievement (Basith et al., 2020; Gopal et al., 2021; Turhan et al., 2022). The mediating effect of students’ satisfaction in the relationship between learning circumstances and academic performance has also been demonstrated (Abd Aziz et al., 2023; Gopal et al., 2021).

Additionally, Study Demands-Resources (SD-R) theories (Lesener et al., 2020; Salmela-Aro et al., 2022a) propose that academic wellbeing mediates the relationship between study demands, such as unpleasant learning circumstances, and academic performance. Specifically, according to the SD-R theories, more demands lead to lower academic wellbeing, i.e. more academic burnout and less academic engagement, which in turn hinders academic performance. Findings of studies conducted during the pandemic indicate that (a higher amount of) ERT, which can be considered as study demand, is associated with more academic burnout and less academic engagement (Harries et al., 2021; Salmela-Aro et al., 2022b; Walker & Koralesky, 2021; Wester et al., 2021; Wissing et al., 2022). It has also been demonstrated that more academic burnout and less academic engagement are predictive of lower academic performance in terms of dropout intentions, grades, and perceived academic achievement (Jiang, 2021; Madigan & Curran, 2021; Romero et al., 2022; Turhan et al., 2022; Vizoso et al., 2018). Finally, the mediating role of academic wellbeing in the relationship between the study demands and academic performance has been confirmed (Marôco et al., 2020; Salanova et al., 2010).

Taken together, there is evidence that students’ study-related experiences, i.e. students’ satisfaction and academic wellbeing, mediate the relationship between characteristics of the university environment and students’ academic performance. However, it is unclear in which sequence students’ satisfaction and academic wellbeing mediate the association between ERT and academic performance. In the occupational context, job satisfaction is theoretically and empirically considered to be an antecedent of work-related wellbeing, i.e. job burnout and job engagement (Rayton & Yalabik, 2014; Shoshani & Eldor, 2016; Zang et al., 2022). The few studies conducted among university students suggest that students’ satisfaction and academic wellbeing are significantly associated (Bélanger & Ratelle, 2021; Bigna et al., 2014; Turhan et al., 2022) and that students’ satisfaction promotes academic wellbeing, i.e. less academic burnout and more academic engagement (Shin & Hwang, 2020; Wissing et al., 2022).

The present study

The aim of the present study was to investigate to what extent the amount of ERT throughout the academic year had a detrimental effect on students’ academic performance in terms of study delay at the end of the academic year and whether study-related experiences throughout the academic year, i.e. students’ education satisfaction and academic wellbeing, are mechanisms underlying the performance-impairing effect of the amount of ERT. To this end, data were collected throughout the academic year 2020/2021 in which ERT was implemented at all conventional Dutch universities as main form of education. The amount of ERT varied throughout the academic year due to changes in the strictness of the preventive measures in force (Universities of The Netherlands, 2022). Additionally, there was a substantial variation in the amount of ERT among students due to differences in the educational offer of the various study programmes, e.g. in periods when limited on campus teaching and learning was possible, priority was given to first-year students and bachelor students (de Boer, 2021).

Based on the presented theoretical models and empirical findings, we expected a higher amount of ERT throughout the academic year to be associated with a higher risk of study delay at the end of the academic year. Additionally, we expected students’ education satisfaction and academic wellbeing throughout the academic year to sequentially mediate this relationship. More specifically, we hypothesised that a higher amount of ERT throughout the academic year is associated with lower education satisfaction throughout the academic year, which in turn is related to lower academic wellbeing throughout the academic year, i.e. more academic burnout and less academic engagement, which in turn increases the risk of study delay at the end of the academic year (see Fig. 1).

Fig. 1
figure 1

Graphic representation of the proposed mediation model

Full size image

Our study extends previous research in two ways. First, the effect of interindividual differences in the amount of ERT throughout a whole academic year was examined while most previous studies investigated the effect of ERT at a group level by comparing periods before and after the onset of the pandemic. Taking interindividual differences into account allows quantifying the risk of experiencing study delay for distinct amounts of ERT. Second, to our knowledge, this is the first study investigating the sequential mediating role of different study-related experiences in the association between study demands and study performance. The insights gained can broaden the understanding of processes underlying the performance-impairing effect of high study demands such as ERT and hereby contribute to the further development of theoretical frameworks and provide practical implications for the prevention of poor study performance.

Materials and methods

Procedure and points of measurement

A detailed description of the procedure, points of measurement, and sampling of the longitudinal online study has been reported elsewhere (Vollmann et al., 2022). The study followed the principles of the Declaration of Helsinki and participants were treated according to the American Psychological Association ethical standards. The Medical Ethics Review Committee of the Erasmus Medical Center has approved the study protocol (#2020–0815). Informed consent was obtained from all participants.

The study was conducted throughout the academic year 2020/2021 that started early in September 2020 and ended mid-July 2021. Data were collected at three time points, i.e. after the first 2.5 months of the academic year (t1, November 18, 2020–December 20, 2020), halfway through the academic year (t2, March 11, 2021–March 28, 2021), and at the end of the academic year (t3, June 28, 2021–July 11, 2021). The first 4 months of the academic year (during t1 and the 2.5 months preceding t1) as well as the last 2.5 months of the academic year (during t3 and the 2 months preceding t3), universities were partly open for on-campus teaching and learning with several restrictions such as 1.5 m distance, wearing a face mask, and a maximum number of students in classes. At t2 and the preceding 2 months, the second national lockdown was in force; universities were fully closed with on-campus teaching and learning being prohibited (except for internships). The average percentage of online education was 87.32% (SD = 19.38), 93.28% (SD = 18.71), and 84.75% (SD = 25.04) during the timespan preceding t1, t2, and t3, respectively (Vollmann et al., 2022).

Full-time students until the age of 30 years old enrolled in a bachelor or master programme at all 13 Dutch conventional universities were eligible to participate. As only a small percentage of students studying at conventional universities is older than 30 years (Statistics Netherlands, 2022), they were excluded from the study. Students were invited via emails sent by two large national student organisations and diverse local student associations to their members as well as via postings on social media of universities, student associations, and study-related organisations. Sampling quotas for university, study phase, gender, and migration background (Statistics Netherlands, 2021) were utilised to obtain a heterogeneous sample.

Participants

A total of 680 students (65.9% female) with a mean age of 21 years (SD = 2.06; range 17–28 years) participated at all three measurement points. Most students were native Dutch (87.2%), while 12.8% had a migration background. Students from all 13 conventional Dutch universities and from all study fields participated. About two-thirds of the students (69.1%) were enrolled in a bachelor programme, while the remainder (30.9%) were enrolled in a master programme.

The present sample is reasonably representative of the Dutch student population in the academic year 2020/2021 regarding university affiliation, field of study, and study phase. However, male students and students with migration background are underrepresented. A detailed description of the (representativeness of the) sample has been reported elsewhere (Vollmann et al., 2022).

Measures

The surveys were administered in Dutch. The amount of ERT, education satisfaction, academic burnout, and academic engagement were measured at all three time points regarding the preceding timespans, i.e. “since September 2020” at t1, “since January 2021” at t2, and “since May 2021 (partial reopening of higher education)” at t3. This way, the measures of the amount of ERT and the study-related experiences span the entire academic year. Study delay was measured at the end of the academic year at t3. This was the most valid time point to determine study delay as the scheduling of resits differs between study programmes and at t3 almost all resits had taken place.

The amount of ERT was measured using a single item, i.e. “What percentage of all your study activities has taken place online?”. This question referred to the share of synchronous (live) online education via a live stream or video calling programme and asynchronous (non-live) online education via recordings or discussion forums vs. traditional synchronous (live) on-campus education at the university. Responses were given as a percentage ranging from 0 to 100%, which were transformed to a 11-point scale ranging from 0 (0%) to 10 (100%) for an easier interpretation of the unstandardised coefficients. The scores at the three time points were averaged to determine the average amount of online education throughout the entire academic year.

Education satisfaction was assessed with the single item “All things considered, how satisfied are you with the educational offer as a whole?” (Sears et al., 2017). Responses were given on a visual analogue scale ranging from 0 (extremely dissatisfied) to 10 (extremely satisfied). The scores at the three time points were averaged to determine the average level of education satisfaction throughout the entire academic year.

Following the theoretical considerations of Kristensen et al. (2005) and recent studies (Lesener et al., 2020), academic burnout was operationalised by its core component emotional exhaustion, which was assessed with the Emotional Exhaustion subscale of the validated Utrecht Burnout Scale for students (UBOS-S) (Schaufeli et al., 2002). The five items (e.g. “I feel emotionally drained by my studies.”) were scored on a 7-point scale ranging from 0 (never) to 6 (always). At each time point, items were averaged so that higher scores indicate more academic burnout. Cronbach’s alpha was ≥ 0.86 at all three time points and evidence for metric measurement invariance across time was found (see electronic supplementary material). The scores at the three time points were averaged to determine the average level of academic burnout throughout the entire academic year.

Academic engagement was measured by using the validated ultra-short Utrecht Work Engagement Scale-Student Form (UWES-3-SF) (Gusy et al., 2019). The three items reflect the dimensions vigour (“When I study, I feel like I am bursting with energy.”), dedication (“My studies inspire me.”), and absorption (“I am immersed in my studies.”). Responses were given on a 7-point scale from 0 (never) to 6 (always). Items were averaged so that higher scores indicate greater academic engagement. Cronbach’s alpha was ≥ 0.76 at all three time points and evidence for metric measurement invariance across time was found (see electronic supplementary material). The scores at the three time points were averaged to determine the average level of study engagement throughout the entire academic year.

Study delay was measured by asking students whether they experienced study delay since the beginning of the current academic year, i.e. whether they have earned less course credits than outlined in their study programme (0 = no, 1 = yes).

Data analyses

Data were analysed using SPSS v28, Mplus v8.8, and Stata v18.0. Firstly, correlations between all study variables were computed to examine their bivariate associations. Subsequently, the hypothesised sequential mediation model (see Fig. 1) with ERT throughout the academic year as independent variable, education satisfaction and academic wellbeing throughout the academic year as serial mediators, and study delay at the end of the academic year as dependent variable was tested with path analysis using logistic regression analysis based on maximum likelihood estimation. The indirect effects of the independent variable on the dependent variable via the mediator(s) were estimated by bootstrapping with 10,000 bootstrap samples (Hayes, 2009). Additionally, to facilitate an easier interpretation of the magnitude of the effect of amount of ERT on study delay, the predicted probabilities of study delay for 11 specific values of amount of ERT (0% to 100% in steps of 10) were calculated using the margins command (Williams, 2012). These adjusted predictions represent the predicted probabilities of experiencing study delay given these specific amounts of ERT.

Results

Bivariate associations between study variables

The results of the correlation analyses (see Table 1) show that a higher amount of ERT throughout the academic year was significantly associated with lower education satisfaction throughout the academic year and a higher risk of study delay at the end of the academic year. Additionally, lower education satisfaction throughout the academic year was significantly related to more academic burnout and less academic engagement throughout the academic year as well as a higher risk of study delay at the end of the academic year. Finally, more academic burnout and less academic engagement throughout the academic year were significantly associated with a higher risk of study delay at the end of the academic year. The same pattern of correlations between the study variables was found at each of the three time points (data not shown).

Table 1 Descriptive statistics of and intercorrelations among study variables
Full size table

Association between amount of ERT and study delay sequentially mediated by education satisfaction and academic wellbeing

The results of the tested sequential mediation model are displayed in Fig. 2. In total, 14.4% of the variance in study delay could be explained by all included variables.

Fig. 2
figure 2

Results of the tested sequential mediation model. Unstandardised coefficients (bs and ORs) are reported. The path coefficient in parentheses represents the total effect. Grey paths are not statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001

Full size image

There was a significant positive total effect of the amount of ERT throughout the academic year on study delay at the end of the academic year with OR = 1.17, 95% CI [1.027, 1.339]. This indicates that for every one-unit increase in amount of ERT (i.e. an increase of 10 percentage points), the odds of experiencing study delay increase by factor 1.17, which is an increase in odds by 17%. The predicted probabilities of study delay for 11 different amounts of ERT are presented in Fig. 3. The predicted probability of study delay was almost four times larger with 100% ERT (31.4% predicted risk of study delay) compared to 0% ERT (8.5% predicted risk of study delay).

Fig. 3
figure 3

Predicted probability (and 95% confidence interval) of study delay at the end of the academic year by amount of ERT throughout the academic year

Full size image

No significant direct effect of the amount of ERT throughout the academic year on the experience of study delay at the end of the academic year was found, OR = 1.13, 95% CI [0.983, 1.324], after adding the study-related experiences throughout the academic year as mediators to the model.

However, three specific indirect effects of the amount of ERT throughout the academic year on study delay at the end of the academic year were significant; first via education satisfaction throughout the academic year, b = 0.031, BC 95% CI [0.011, 0.061]; second via education satisfaction throughout the academic year and subsequently academic burnout throughout the academic year, b = 0.012, BC 95% CI [0.005, 0.023]; and third via education satisfaction throughout the academic year and subsequently academic engagement throughout the academic year, b = 0.014, BC 95% CI [0.005, 0.028]. These indirect effects indicate that a higher amount of ERT was associated with an increased risk of experiencing study delay through lower education satisfaction as well as through lower education satisfaction and subsequently more academic burnout and less academic engagement.

Additionally, indirect effects of the amount of ERT throughout the academic year on academic burnout throughout the academic year, b = 0.039, BC 95% CI [0.020, 0.061], and on academic engagement throughout the academic year, b =  − 0.038, BC 95% CI [− 0.054, − 0.020], via education satisfaction throughout the academic year were significant. These effects indicate that a higher amount of ERT was associated with more academic burnout and less academic engagement through lower education satisfaction.

The results did not change after controlling for study phase, gender, and migration background (data not shown).

Discussion

The present study investigated the association between the amount of ERT and study delay as well as explored education satisfaction and academic wellbeing as sequential mediators in this association among university students in the Netherlands.

As expected, we found a higher amount of ERT throughout the academic year to be associated with a higher risk of experiencing study delay at the end of the academic year. The risk of study delay was almost quadrupled when all educational activities took place online compared to when no educational activities took place online. This finding adds to previous research detecting a decline in students’ academic performance after the onset of the corona crises and the introduction of ERT (Aucejo et al., 2020; Bird et al., 2022; Farcnik et al., 2021; Interstedelijk Studenten Overleg, 2020; Jiang, 2021; Kim et al., 2021; Mahdy, 2020; Oducado & Estoque, 2021; Tinjić & Halilić, 2020; van den Broek et al., 2022), by demonstrating that also the amount of online educational activities in relation to on-campus teaching and learning (even with a number of restrictions) matters when it comes to academic performance.

The present study further extends previous research by investigating study-related experiences, i.e. educations satisfaction and academic wellbeing, as mechanism underling the performance-impairing effect of ERT, with a special focus on the sequential order of these experiences. In accordance with several theoretical frameworks (Duque, 2014; Lesener et al., 2020; Salmela-Aro et al., 2022a; Schertzer & Schertzer, 2004), both education satisfaction and academic wellbeing served as mediators in the association between the amount of ERT and study delay. Education satisfaction seems to play a central role in this mediation process. A higher amount of ERT throughout the academic year was particularly related to lower education satisfaction throughout the academic year, which in turn increased the risk of study delay at the end of the academic year directly as well as indirectly through being associated with lower academic wellbeing, i.e. more academic burnout and less academic engagement, throughout the academic year.

Interestingly, in the present study, no significant association was found between the amount of ERT and students’ academic burnout and engagement, which is contrary to the assumptions of the SD-R theories (Lesener et al., 2020; Salmela-Aro et al., 2022a) and previous studies demonstrating a decline in students’ academic wellbeing due to ERT (Daniels et al., 2021; Harries et al., 2021; Pasion et al., 2021; Salmela-Aro et al., 2022b; Vollmann et al., 2022; Wester et al., 2021; Wissing et al., 2022). However, other previous studies found students’ academic wellbeing to be unaffected (Pasion et al., 2021; Schindler et al., 2021; Zis et al., 2021) or even positively affected (Alkureishi et al., 2022; Chew et al., 2021; Goppert & Pfost, 2021) by ERT. These mixed findings suggest that there might be other factors, such as students’ sociodemographic and individual characteristics (e.g. gender, study progress, personality, competences, and social integration), that alter the effect of (amount of) ERT on academic wellbeing. For example, it has been found that the effect of ERT on the academic burnout dimension exhaustion was strongest among medical students in their last clinical year (Zis et al., 2021) and that ERT had a positive effect on academic engagement only in students with high technology self-efficacy (Owusu-Agyeman et al., 2021). On the other hand, we found the amount of ERT to be indirectly associated with lower academic wellbeing via lower education satisfaction, which is in accordance with earlier findings from the educational and occupational context (Rayton & Yalabik, 2014; Wissing et al., 2022; Zang et al., 2022). This pattern of results, i.e. absence of a bivariate association of amount of ERT with academic wellbeing and presence of an indirect association between these variables via education satisfaction, indicates a temporally distal process which is transmitted through an additional variable (Shrout & Bolger, 2002). Thus, it can be assumed that the proposed sequence of job satisfaction as antecedent of work-related wellbeing in the occupational context (Rayton & Yalabik, 2014; Shoshani & Eldor, 2016; Zang et al., 2022) also applies to the academic context in which education satisfaction precedes academic wellbeing, i.e. academic burnout and academic engagement. Overall, this finding again underpins the central role of education satisfaction in the mechanism underlying the detrimental effect of ERT on academic performance.

It should be noted that the large majority of the students in our sample did not experience study delay in the academic year 2020/2021. However, no less than 27.8% of the students earned less course credits than outlined in their study programme, which concurs with the finding of the Dutch Inspectorate of Education (2021) that 27.6% of the university students reported experiencing study delay due to the corona crisis. Unfortunately, official long-term registers record study delay not regarding an academic year but the entire study period (Ministry of Education Culture and Science, 2021; van den Broek et al., 2022), which makes a comparison of our rates during and the recorded rates before the pandemic impossible.

Finally, only 14.4% of the variance in study delay could be explained by our model. However, this is not surprising as variables that are known to be substantial predictors of academic performance, e.g. past performance, scholastic aptitude, goal setting, and self-efficacy (Richardson et al., 2012), were not included. Nevertheless, the present study gives inside in the processes that underlie the performance-impairing effect of ERT.

Practical implications

Since high levels of ERT increase the risk of study delay, policymakers and universities should carefully deliberate a (partial) closure of university campuses in future waves of COVID-19 or other disastrous events. When ERT must be implemented due to severe public health concerns or safety issues, it should be offered in such a way that students’ satisfaction is preserved in order to protect their academic wellbeing and consequently their academic performance. The same might be true for the implementation of online education as part of the regular university curriculum. Previous research revealed several determinants of students’ satisfaction in the context of ERT and regular online education, with social interactions and attributes of the online learning environment revealing to be most crucial (Agyeiwaah et al., 2022; Eom & Ashill, 2016; Gashi et al., 2022; Harsasi & Sutaeijaya, 2018; Nyathi & Sibanda, 2022; Parahoo et al., 2016; Soliman et al., 2022). As meaningful and supportive interactions with both teachers and peers promote students’ satisfaction (Eom & Ashill, 2016; Gashi et al., 2022; Nyathi & Sibanda, 2022; Parahoo et al., 2016; Soliman et al., 2022), plenty of opportunities for synchronous and asynchronous collaboration should be provided during ERT and other forms of online education. Also, previous research showed that students are more satisfied with any form of online education when the online environment and material is stimulating and attractive; when the technology is easy to use, functional, and reliable; and when the course is well structured, but also allows a certain flexibility (Agyeiwaah et al., 2022; Eom & Ashill, 2016; Gashi et al., 2022; Harsasi & Sutaeijaya, 2018; Soliman et al., 2022). Universities are advised to invest in developing and updating (online) education, in reliable and user-friendly online tools, and in professionalising teachers so they can make optimal use of all innovative (digital) teaching possibilities.

Limitations

Some limitations of the present study need to be mentioned. To begin with, although the study had a longitudinal design, the tested model includes causal paths that are based on cross-sectional data, i.e. the effects of education satisfaction throughout the academic year on academic wellbeing throughout the academic year. The proposed temporal order of the variables is derived from theoretical assumptions and previous empirical findings; however, with cross-sectional data, the direction of causality cannot be conclusively determined (Winer et al., 2016). Additionally, there are three issues regarding the measured variables. First, information on the amount of ERT and study delay was based on self-reports instead of administrative data and might therefore be subject to biases. Recall bias regarding the amount of ERT was reduced by asking students various times during the academic year about the past 2.5 to 3 months. Self-reports of academic performance reasonably reflect the actual performance in students with high ability and good grade point averages but could be positively biased in students with lower grade point averages and lower ability (Kuncel et al., 2005). Second, only the core component of burnout, i.e. emotional exhaustion, was investigated. To gain a comprehensive understanding of the antecedents and consequences of academic burnout, the components cynicism and academic inefficacy require future investigation. Third, this study focused on study delay as indicator of academic performance. No conclusions can be drawn regarding other performance measures such as grade point average or attrition. Finally, although the sample was largely representative of the Dutch student population, male students and students with migration background were underrepresented. Moreover, due to using the convenience sampling method, information on response rate and non-responder characteristics is unavailable. Also, there might be a selection bias towards highly resilient and motivated students.

Conclusions

The amount of ERT during the COVID-19 pandemic was associated with an increased risk of study delay among students studying at conventional Dutch universities. The central mechanism underlying this effect appeared to be students’ dissatisfaction with the educational offer. More ERT was related to lower education satisfaction, which in turn increased the risk of study delay directly as well as indirectly through lower academic wellbeing, i.e. more academic burnout and less academic engagement. ERT, but also regular online education should be offered in such a way that students’ satisfaction is preserved in order to protect their academic wellbeing and consequently their academic performance.