Korean society has high aspirations for academic achievement, with excessive value placed on studying. This phenomenon stems from Korea’s rapid economic development. Education is considered an essential element for keeping pace with this swift growth, and admission to prestigious universities is viewed as a prerequisite for attaining high social status. The competition for entrance to prestigious universities has become increasingly arduous (M. Lee & Larson, Reference Lee and Larson2000). Owing to this rigorous climate, teachers and parents put much academic pressure on students. In turn, this causes conflicts between parents and their children, leading to undesirable levels of academic motivation and academic stress (Bong et al., Reference Bong, Kim, Shin, Lee and Lee2008). This scholastic context may be a reason for the incongruity between Korean students’ high academic performance and poor psychological states. According to results from the Program for International Student Assessment (PISA), Korean students had much higher academic achievement scores than the average among the Organisation for Economic Cooperation and Development (OECD) countries. However, psychological elements such as joy, interest, motivation, and self-efficacy were lower than the average of OECD countries (Korea Institute for Curriculum and Evaluation, 2016). In addition, most Korean students feel perturbed because of their studies. According to the National Youth Policy Institute (Kim & Yu, Reference Kim and Yu2016), 68.9% of Korean middle school students and 80% of high school students reported experiencing academic stress. Consequently, they gradually become exhausted and experience burnout as they get older (J. Lee et al., Reference Lee, Puig, Lea and Lee2013). Notably, burnout is a psychological symptom of chronic stress.

Academic burnout studies take the concept of the job burnout model and apply it to a school context. This is an apt comparison, since academic activities (such as attending classes and doing homework) can be considered tasks (Schaufeli & Taris, Reference Schaufeli and Taris2005). Academic burnout consists of three subdimensions: exhaustion, cynicism, and inefficacy. Exhaustion refers to emotional and physical fatigue. Cynicism is an expression of indifferent and distant attitudes towards academic activities. Inefficacy refers to a lack of academic accomplishment (Schaufeli, Martinez et al., Reference Schaufeli, Martinez, Pinto, Salanova and Bakker2002). Among these three constructs, exhaustion has been referred to as the core factor in burnout (Schaufeli & Taris, Reference Schaufeli and Taris2014). When students perceive that their exhaustion results from academic stress, they maintain distance from or have negative feelings towards academic activities. In so doing, they can conserve energy. Therefore, many scholars (e.g., Kristensen et al., Reference Kristensen, Borritz, Villadsen and Christensen2005; Shirom, Reference Shirom2003) have suggested narrowing the concept of burnout down to exhaustion. The exhaustion items on the Maslach Burnout Scale (MBI) have frequently been used as a measure of burnout (Bekker et al., Reference Bekker, Croon and Bressers2005). Accordingly, this study used this exhaustion scale to measure academic burnout.

Engagement is defined as energy, involvement, and efficacy (a concept which is the opposite of burnout; Maslach et al., Reference Maslach, Schaufeli and Leiter2001). Engagement has three components: vigour, dedication, and absorption. Individuals with high levels of engagement have intrinsic motivation and good feelings toward their job (Schaufeli, Martinez et al., Reference Schaufeli and Salanova2002). Vigour means having elevated levels of energy and a willingness to devote that energy to one’s work while maintaining mental resilience. Dedication involves deriving enthusiasm, a sense of challenge, and meaning from one’s work. Absorption refers to being happily engrossed in one’s work and hesitating to part from work. Among the three constructs, vigour and dedication have been recognised as the core dimensions of engagement (Schaufeli & Bakker, Reference Schaufeli and Bakker2004). However, only vigour is significantly related to academic achievement in school settings (Casuso-Holgado et al., Reference Casuso-Holgado, Cuesta-Vargas, Moreno-Morales, Labajos-Manzanares, Barón-López and Vega-Cuesta2013; Schaufeli, Martinez et al., Reference Schaufeli, Martinez, Pinto, Salanova and Bakker2002). Moreover, vigour is the conceptual opposite of exhaustion (Maslach & Leiter, Reference Maslach and Leiter1997; Schaufeli, Salanova et al., Reference Schaufeli, Martinez, Pinto, Salanova and Bakker2002). Thus, in this study, only vigour was used to measure academic engagement.

Currently, the revised job demand-resources (JD-R) model (Schaufeli & Bakker, Reference Schaufeli and Bakker2004) is the most popular model for describing both burnout and engagement. The JD-R model is characterised by two different underlying psychological processes (i.e., the health impairment and motivational processes). Each plays a role in the development of job-related strain and motivation. The health impairment process explains how burnout has mediation effects on the relationship between continuous job demands and health problems such as depression and cardiovascular disease (Schaufeli & Taris, Reference Schaufeli and Taris2014). The motivational process refers to flourishing job resources (which motivate engagement) and their impact on positive work outcomes. In previous research (Demerouti et al., Reference Demerouti, Bakker, Nachreiner and Schaufeli2001), the JD-R model assumed that job demands are related to feelings of exhaustion, whereas a lack of resources is related to disengagement from work. Typical examples of job demands are physical workloads, time pressure, recipient contact, physical environments, and shift work. Meanwhile, examples of job resources are positive feedback, rewards, job control, job security, and supervisor support. In the JD-R model, characteristics of these contributing factors are not limited to specific job demands or job resources. Accordingly, it has been confirmed that academic burnout and academic engagement can be explained by adopting the JD-R model (Cho et al., Reference Cho, Lee, Lee and Lee2018; Moon et al., Reference Moon, Kim and Lee2018; Salmela-Aro & Upadyaya, Reference Salmela-Aro and Upadyaya2014).

Academic demand (namely, academic pressure) means that students are required to undertake physical and psychological efforts and costs to achieve academic success (Bakker & Demerouti, Reference Bakker and Demerouti2007). For instance, students must perform overwhelming academic tasks in limited time periods and competitive situations. In this context, intense class climates could be considered important and stressful environments for students. In competitive classrooms, teachers force their instructional agendas on their students, thus hampering the students’ inner motivation and engagement. This is problematic, in that these curriculums do not match students’ needs and interests (Reeve et al., Reference Reeve, Jang, Carrell, Jeon and Barch2004). As a result, students show lower engagement levels when paired with oppressive teachers. However, students with teachers supportive of autonomous academic activities perform more enthusiastically and tend to display better psychological states (O’Connor & Vallerand, Reference O’Connor and Vallerand1994). In this study, teachers’ academic pressure was set as the criterion for dividing groups.

Social support is the most well-known and effective resource in work contexts (Cohen & Willis, Reference Cohen and Wills1985; Bakker et al., Reference Bakker, Demerouti, Taris, Schaufeli and Schreurs2003). High-quality social support from co-workers and supervisors buffers the negative effects of workload and stress on burnout (Bakker et al., Reference Bakker, Demerouti and Euwema2005; Bakker et al., Reference Bakker, Hakanen, Demerouti and Xanthopoulou2007). Moreover, social support has proven to have a positive effect on engagement (Garcia-Reid et al., Reference Garcia-Reid, Reid and Peterson2005; Wang & Eccles, Reference Wang and Eccles2012). In the school context, especially for adolescents, peer support is more important than other resources. This is attributed to the fact that during puberty, self-concepts are developed through relationships with other students (Choi, Reference Choi2018). According to previous research (Mead et al., Reference Mead, Hilton and Curtis2001), peer support is ‘a system of giving and receiving help founded on key principles of respect, shared responsibility, and mutual agreement of what is helpful’ (p. 6). In addition, social support is based on the empathic understanding of others’ situations via sharing painful emotional and psychological experiences. Moreover, peer support has positive effects on the development of adolescents’ psychological wellbeing and self-identity (Buchanan & Bowen, Reference Buchanan and Bowen2008; Kaplan et al., Reference Kaplan, Cassel and Gore1977).

Interestingly, the effects of resources such as social support are more salient under stressful situations than in situations with low levels of stress. According to the conservation of resources (COR) theory (Hobfoll, Reference Hobfoll1989, Reference Hobfoll2002), people mobilise and use their resources in stressful situations to prevent burnout. In dealing with stressful situations, therefore, people need increased access to supportive social structures. When a situation is not so demanding that it threatens one’s valued resources, they would not need to deploy them. In research on information services workers, the effects of personal resources on burnout (e.g., self-efficacy and coping style) were salient when work resources (e.g., supervisor support) were low. However, when work resources were high, personal resources were not beneficial in preventing workers’ burnout (Riolli & Savicki, Reference Riolli and Savicki2003). Bakker et al. (Reference Bakker, Hakanen, Demerouti and Xanthopoulou2007) found that the relationship between the use of job resources and teachers’ work engagement was stronger when teachers were faced with significant pupil misbehaviour. (Control of students’ misbehaviour was seen as teachers’ main job requirement.) As is evidenced by these research results and COR theory, it is hypothesised that the effects of academic resources could vary, based on demand levels.

Meanwhile, Schaufeli and Taris (Reference Schaufeli and Taris2014) pointed out the importance of psychological mediators in explaining the mechanism of relationships between demand-resources and burnout-engagement. Furthermore, they suggested that it is necessary to integrate personal vulnerability factors (e.g., neuroticism, workaholism, and pessimism) to extend the JD-R model, as thus far only personal resources have frequently been studied in JD-R research. In this study, we set ‘academic hatred’, which is synonymous with hatred of academic works (Lee et al., Reference Lee, Cho and Lee.2019), as a personal vulnerability factor to describe why students would experience burnout or be less engaged with their studies. Although this concept was first introduced as a new dimension of the Korean Academic Burnout Inventory (KABI; Y.B. Lee et al., Reference Lee, Lee and Lee2009) and labelled as antipathy, it has not been considered in burnout research. Furthermore, most of the items that measure antipathy in KABI are measured with the predicate ‘hate’, so some researchers (M. Lee, Reference Lee2020; M. Lee et al., Reference Lee, Lee and Lee2020) questioned the term ‘antipathy’ and tried to elaborate the concept of academic hatred by applying the typical characteristics of ‘hate’ to the academic context. ‘Hate’ is a secondary emotion that combines emotion and cognitive evaluation and has the characteristic of an emotional attitude that can persist beyond simple emotion (Ekman, Reference Ekman1992). In previous studies (Fischer et al., Reference Fischer, Halperin, Canetti and Jasini2018; Sternberg, Reference Sternberg and Sternberg2005), the emotion of hate reflects the individual’s cognitive evaluation of the anger or disgust and the context in which these primary (or basic) emotions are triggered. Moreover, it is accompanied by a sense of powerlessness in the inability to change the threatening situation.

According to Lee (Reference Lee2020), academic hatred can be defined as a hostile emotional attitude towards academic works. This concept is based on the control-value theory (Pekrun, 2000) and reflects the Korea’s competitive academic context, the emotions students experience in this context, and the individual’s cognitive evaluation of these. It is based on the theoretical view that various academic emotions can be caused depending on how much students perceive their control of their academic context and whether they value academic tasks and outcomes and themselves as learners (M. Lee et al., Reference Lee, Lee and Lee2020). In conclusion, students who feel physical or psychological threats owing to their studies would feel powerless when realising that they can no longer change their negative and threatening academic situation on their own, thus leading them to experience academic hatred (M. Lee, Reference Lee2020; M. Lee et al., Reference Lee, Lee and Lee2020).

Students with academic hatred express hostility towards their studies or environment, and thus would disengage from them. As academic hatred is one dimension of emotions, intensity and frequency of this feeling could be affected by both one’s characteristics and their context (Pekrun, Reference Pekrun2006). According to the results of multilevel analysis performed by M. Lee et al. (Reference Lee, Cho and Lee.2019), at the individual (intrapersonal) level, academic hatred was affected by depression, test anxiety, and the amount of academic pressure provided by parents. Lower socioeconomic status and higher level of teachers’ support for students’ autonomy engendered lower academic hatred at class (environmental) level. These research results showed that academic hatred is influenced by both environmental and individual characteristics. Many Korean teachers and parents force students to view their peers as competitors and exhort them to win competitions. Considering that these contextual factors stand out in Korea’s academic climate, hatred should be considered in the study of academic burnout and engagement among Korean students.

This study investigated the effects of teacher-originated academic pressure, peer support, and academic hatred on the processes of academic burnout and engagement based on the JD-R model. Moreover, to evaluate the interaction effects between academic demand and resources, we conducted multigroup structural equation model (SEM) analysis. The participants of this study were senior high school students who were facing Korea’s national college entrance exam in a month’s time. Under these circumstances, their academic stress would be at its highest (E.H. Lee, Reference Lee2009). Finally, only the dimensions of exhaustion and vigour were used, as they are the core factors of academic burnout and engagement respectively (Schaufeli & Taris Reference Schaufeli and Taris2014).

Methods

Participants and Procedures

This study used data from final-year high school students who were taking Soo-Neung (Korean College Scholastic Ability Test, or KCSAT) within a month’s time, when students experience the highest levels of academic stress. For representative sampling, demographic characteristics (e.g., GPA, socioeconomic status, gender) were considered. As a result, eight schools, including 43 classes (on academic-only tracks), were assessed in this research. A total of 1015 students completed the survey, of whom 582 (57.3%) were female, with a response rate of 94.2%. Table 1 shows demographics and mean values (i.e., percentage of females, percentage of students of lower socioeconomic status, and number of classes). It also indicates mean scores of teachers’ academic pressure at class levels among the two groups (in either competitive or noncompetitive classroom climates).

Table 1.

Demographics and mean values of two groups

Note: LP, Low level of teacher’s academic pressure group; HP, High level of teacher’s academic pressure group.

All participants agreed to provide personal information (e.g., gender), and voluntarily completed paper questionnaires under researchers’ instructions. This study received ethical clearance from the institutional review board at one of South Korea’s universities. To deal with missing data, we evaluated missing data completely at random (MCAR) by expectation-maximisation estimated statistics. The result of Little’s MCAR test was accepted by the null hypothesis (χ²=296.826, p = .247). This means that there is no suspicious pattern in, or distribution of, the missing data. Hence, full information maximum likelihood (FIML) was conducted to treat the missing data.

Measures

Teacher academic pressure

This study used the survey items from the Korean Educational Longitudinal Study (KELS, 2005), conducted by the Korean Educational Development Institute (KEDI), to measure teachers’ pressures for academic achievement. It comprises six items on a 5-point, Likert-type scale (e.g., ‘My teacher emphasises high academic performance in examinations’. The reliability for teacher pressure was .71 in a previous study (Jeong, Reference Jeong2015), and .78 in this study.

Peer support

The Social Support Inventory (SSI) is a scale which was cross-validated by M.S. Kim (Reference Kim1995) using the original version of the Social Support Scale for Children (SSSC; Harter, 1985) and based on Korean academic culture. It includes three subscales: parental support, teacher support, and peer support. In this study, we used only the peer support questionnaire to measure social support by friends. It comprises six items on a 5-point, Likert-type scale (e.g., ‘My friends are supportive of what I do’). The reliability for peer support was .71 in M.S. Kim (Reference Kim1995), and .91 in this study.

Academic Burnout and Academic Hatred

We used the KABI (Lee, Reference Lee2009) to measure exhaustion and academic hatred. The KABI is based on the Maslach Burnout Scale–Student Survey (MBI-SS; Schaufeli, Martinez et al., Reference Schaufeli, Martinez, Pinto, Salanova and Bakker2002), which consists of three subscales (i.e., exhaustion, cynicism, and inefficacy). In addition, it captures new psychological characteristics (i.e., antipathy and anxiety) among Korean students, and added them as components of academic burnout. It was validated by Y.B. Lee et al. (Reference Lee, Lee and Lee2009). All scales are 5-point, Likert-type scales, each consisting of five items. In this study, only two dimensions were utilised: exhaustion (e.g., ‘I feel emotionally drained by my studies’) and antipathy (e.g., ‘I hate to study’). Antipathy items corresponded to, and were used for, assessment of academic hatred because (despite its importance) no direct measures have been developed. Specifically, the five items for antipathy include: ‘I hate to study’, ‘I hate a world that focuses only on studying’, ‘I hate a world that only prioritises studying’, ‘I want to live in the world without studying’, and ‘I am irritated with studying’. The reliability coefficients for exhaustion and antipathy were .85 and .90 in the original study, and .89 and .89 in this study respectively.

Academic engagement

The Korean Academic Engagement Inventory (KAEI) was developed and validated by J. Lee and Lee (Reference Lee and Lee2012). This measure was based on the Utrecht Work Engagement Scale-Students (UWES-S; Schaufeli, Martinez et al., Reference Schaufeli, Martinez, Pinto, Salanova and Bakker2002), and was applied in the current study to assess students’ academic engagement. The KAEI consists of four subscales (i.e., vigour, dedication, absorption, and efficacy), and each dimension contains four items with 5-point, Likert-type scale. A typical item for measuring vigour is ‘When studying I feel strong and vigorous’. The reliability coefficients for vigour were .89 in both the original study and this study.

Statistical Analysis

In this study, we used the multigroup mediation model to examine the different effects of classroom climates (which are formed by high and low levels of teachers’ academic pressure). SPSS 21.0 was used in descriptive statistics and intercorrelations analysis, while AMOS 21.0 was used to run the multigroup mediation model. Teachers’ academic pressure was calculated to the average score of each class, and converted into z-scores (i.e., standardised scores). The z-scores with positive numbers were included in the group (competitive climate group) with high levels of teachers’ academic pressure. The negative z-score group was constructed as a group (noncompetitive climate group) with low levels of teachers’ academic pressure. The mean split was conducted in this study based on previous research for examining multigroup analysis in vocational and burnout studies (B. Kim et al., Reference Kim, Lee, Ha, Yang and Lee2016; Slaney & Brown, Reference Slaney and Brown1983).

Three fit indices were used in this paper to evaluate models; namely, the root-mean-square error of approximation (RMSEA) with the comparative fit index (CFI) and Tucker-Lewis index (TLI). The chi-square testing is sensitive to sample size; thus, it was not adopted as a fit index in this research. The models with CFI and TLI above .90 and RMSEA of below .06 are considered good models. Scaled chi-square difference and ΔCFI were used to compare models. The models differ in that the difference of scaled chi-square is significant, and the ΔCFI is greater than .01. Bootstrap analysis was run to estimate indirect effects with bias-corrected standard errors and 95% confidence intervals (CI).

Results

Descriptive Analysis and Measurement Models

Table 2 demonstrates means, standard deviations, comparison of means and correlation matrix of both the low and high teacher pressure groups. A one-way analysis of variance (ANOVA) was performed to investigate whether there are differences in the means of variables between the two groups. The results demonstrated significant differences between means of the variables of the two groups. The students in the high teacher pressure group (competitive classroom) tended to perceive higher levels of peer support, show higher rates of exhaustion, and have higher academic vigour than students in the low teacher pressure group (non-competitive classroom). However, the mean of academic hatred was higher for the low teacher pressure group than for the high teacher pressure group.

Table 2.

Descriptive statistics and intercorrelations of the research variables for LP (n = 557; below the diagonal) and HP (n = 458; above the diagonal)

Note: **p < .001, ^** p < .01, ^* p < .05.

Structural Models

Group differences in the relationship between latent variables can be examined only if they are perceived identically in all groups (Hong et al., Reference Hong, Malik and Lee2003). As recommended by Steenkamp and Baumgartner (Reference Steenkamp and Baumgartner1998), configural invariance and path invariance were evaluated in hierarchical order. To investigate whether the measured variables were identically perceived across high and low teacher pressure groups, configural invariance was examined. If no significant difference existed between the unconstrained model and the constrained model, configural invariance would be proved. It did not show significant difference in ΔCFI (.002) in comparison with the unconstrained model; the model with factor loading was equally constrained across groups. To evaluate path invariance, the factor loading constrained model was compared with the path coefficients constrained model. There was no significant difference in ΔCFI (.000), suggesting that the model met path invariance.

For the next step, the direct effects of peer support on each variable were evaluated: academic hatred, academic exhaustion, and academic vigour. The direct effects of peer support on academic hatred and exhaustion were not significant in the group with low teacher pressure (academic hatred: β = −.09, p = .58; exhaustion: β = −.08, p = .62), but were significant in the group with high teacher pressure (academic hatred: β = −.20, p < .001; exhaustion: β = −.17, p < .001). The results support H1 (differential effects of peer support on exhaustion), as peer support was negatively related to exhaustion only in students with high teacher pressure. No significant relationship was found between peer support and exhaustion in the group with low teacher pressure. The data from both groups presented the direct effect of peer support on vigour (high teacher pressure group: β = .17, p < .001; low teacher pressure group: β = .18, p < .001). Therefore, H2 (differential effects of peer support on vigour) was not supported. This showed that peer support was positively related to vigour in students with both high and low teacher pressure. However, the results support H3 (differential effects of peer support on academic hatred), as peer support was negatively related to academic hatred only in students with high teacher pressure. In the low teacher pressure group, an insignificant relationship was found between peer support and academic hatred. Academic hatred was positively related to academic burnout in both samples (high teacher pressure group: β = .48, p < .001; low teacher pressure group: β = .32, p < .001) and negatively related with academic vigour (high teacher pressure group: β = −.55, p < .001; low teacher pressure group: β = −.48, p < .001).

As shown in Figure 1, we evaluated the mediation effect of academic hatred in the relationship between peer support and academic exhaustion using multigroup SEM models (high teacher pressure group vs. low teacher pressure group). To investigate the indirect effects of peer support, bias-corrected bootstrapping was performed. In the high teacher pressure group, the bootstrap estimate of the mediation effect from peer support to exhaustion through academic hatred was β = −.095 and did not contain zero on a 95% confidence interval [−.16, −.05]. However, there was no significant mediation effect among the group of low teacher pressure β = −.028, CI [−.07, .00]. This shows that academic hatred has a mediation effect in the relationship between peer support and academic exhaustion on the group in high teacher pressure climates. The same cannot be said of the group under low teacher pressure. The examination of the mediation effect of academic hatred on the relationship between peer support and academic vigour was conducted in the same way. There were significant mediation effects on the group of high teacher pressure, β = .127 CI [.06, .21], whereas no significant mediation effects on the group of low teacher pressure β = .046 CI [−.01, .11] were present. Thus, the mediation effect of academic hatred was significant only in the group with high teacher pressure. Therefore, H4 (differential mediation effects of academic hatred in the relationship between peer support and academic burnout/academic engagement) was supported.

Figure 1.

Final model (Standardised path coefficients) for both groups.

Note: Values in brackets correspond to the results of the low level of teacher’s academic pressure group.