Self-efficacy beliefs are central to Social Cognitive Theory (Bandura, 1997) due to their effects on human functioning, and cognitive, affective, motivational and selective processes. People create and develop judgments of ability that assist them in setting goals and the partial control they can exert over their environment, as research across a range of cultural contexts demonstrated (Bandura, 1977, 1997). In the educational field, self-efficacy has gained prominence in teaching and learning from primary education to graduate studies, including the work and career development of teachers and lecturers.

Lecturer self-efficacy is defined as “the judgments that lecturers make about their ability to teach, research, complete extension tasks and carry out management activities at a level of quality appropriate to their institution’s needs” (Matos et al., 2020, p. 3). Because people decide to act according to perceptions of their own abilities, self-efficacy beliefs influence goal setting, the selection of favorable environments, the efforts made to achieve goals, and even the levels of physiological activation encountered in stressful situations (Bandura, 2012). Consequently, self-efficacy can directly affect a lecturer’s performance and well-being at work. Research conducted over past decades has demonstrated that lecturer self-efficacy impacts upon different psychological processes, including motivation (Bailey, 1999), positive and negative affect (Burton et al., 2005), job satisfaction (Mottet et al., 2004; Ismayilova and Klassen, 2019), self-esteem (Evans and Tress, 2009), and emotional intelligence (Ali et al., 2017), among others.

In higher education, self-efficacy has been associated with lecturer burnout by many researchers, with a role as protective of health (Evers et al., 2002; Skaalvik and Skaalvik, 2010; Savas et al., 2014; García Padilla et al., 2017; Smetackova, 2017; Cao et al., 2018; Llorca-Pellicer et al., 2021). The negative relationship between self-efficacy and burnout is to be theoretically expected since burnout can be understood as a consequence of an inability to deal with contextual demands and workload stress. As strong self-efficacy beliefs empower individuals to deal with their work-related conditions, lecturers with a stronger sense of self-efficacy than others may feel more fulfilled and more satisfied with their work, and present with lower levels of exhaustion and depersonalization (Morris et al., 2017).

Self-efficacy may also have a high predictive value in positive factors associated with lecturer well-being, since self-efficacy can help lecturers to stay motivated and satisfied (Zee and Koomen, 2016). Faced with a lack of research evidence, however, the relationship between lecturer self-efficacy and quality of life is not entirely clear, presenting a knowledge gap which we begin to address here. In schools, however, teacher self-efficacy has been shown to have a positive and significant association with quality of life in Iran (Shirazi et al., 2008). It has also been shown that teacher’s coping self-efficacy in school mediates the relationship between violence-related stress and quality of life (Won and Chang, 2020).

Bandura (1997) states that if lecturers can control the way they react to stressful situations, then they are probably more capable of dealing with them more effectively. Thus, lectures who believe they can manage difficult or challenging work-related events are less likely to be distressed by them. On the other hand, those who believe otherwise might experience higher levels of concern. Teachers with higher self-efficacy scores tend to develop proactive behaviors for facing professional challenges and emotional stress (Yin et al., 2020). Self-efficacy is also positively associated with higher well-being (Song, 2021). Considering work-related studies across a range of fields, self-efficacy has been associated with other adverse aspects of well-being and performance in addition to burnout (Judge and Bono, 2001).

As Bandura (1997) points out, work assumes a central role in the lives of lecturers, acting as a source of personal identity, self-worth and social relationships. We predict that lecturers with higher levels of self-efficacy beliefs should therefore have a better perception of their quality of life, since their self-efficacy beliefs should influence their ability to cope with the contextual demands of work, decreasing the impact of negativity on their health and well-being (see also Skaalvik and Skaalvik, 2010; Sariçam and Sakiz, 2014; Zee and Koomen, 2016; Kim and Burić, 2019). If lecturer self-efficacy acts as a predictor of quality of life and protective against burnout, supporting lecturers more effectively should be a priority for universities as self-efficacy enhancement would also be investing in staff development and well-being. Perceptions of quality of life also involve an appraisal of personal and contextual variables which different individuals deal with every day, including gender, personal relationships, and leisure time activities, which also affect appraisals of self-efficacy and burnout (Cao et al., 2018; Matos et al., 2021). These assumptions and the variables outlined sit at the core of our work and which we explore through structural equation modeling (SEM) and cluster analysis. Two research questions focus attention and guide progress:

We addressed the research questions employing a correlational, cross-sectional design involving the use of an online questionnaire survey method.

We adopted a non-probabilistic and convenience sampling approach in which the participants were self-selecting and voluntary. The only inclusion criterion for participants was to be actively working as a higher education lecturer in the Brazilian public and/or private sector. After obtaining ethical approval from the lead institution, we carried out the data collection online. Doing so, we disseminated a survey link by e-mail to the human resources departments of all 199 universities listed in the 2017 census of Brazilian higher education, asking them to forward the invitation to participate on to lecturers themselves. We also adopted the strategy of sending the link directly to lecturer email addresses where these were available on university websites.

By accessing the survey link, participants were directed to a consent form on Google Forms. Upon completion, they were then directed to the data collection instruments that were made available on the OnlinePesquisa platform. Data collection took place anonymously between October 2019 and January 2020.

The participants recruited to the study were 1,709 lecturers in 78 public (96.1%) and private (3.9%) universities in 26 states across all regions and the Federal District (Table 1). Data obtained from the 2017 census of higher education indicated that the sample was drawn from 124,291 Brazilian lecturing positions overall. In terms of background characteristics, men and women were almost equally represented (51.9% female). Most participants were also white (78.3%), married (72.9%), qualified to doctoral level (89.4%) and between 30 and 49 years of age (63.9%).

The data collection instruments included the Brazilian Lecturer Self-Efficacy Scale (BLSES), the Maslach Burnout Inventory (MBI-ES), and the World Health Organization Quality of Life Assessment (WHOQOL-Bref). We also collected additional demographic information using a questionnaire separately as indicated (Table 1).

The Brazilian Lecturer Self-Efficacy Scale (BLSES) specifically developed for use in this research (Matos et al., 2020) is a 30-item questionnaire adopting a 5-point Likert scale. The 30 items are arranged into four factors: self-efficacy for teaching activities (α = 0.86), self-efficacy for research activities (α = 0.87), self-efficacy for extension activities (α = 0.86), and self-efficacy for university management activities (α = 0.85), each reflecting the main duties of lecturers in Brazil. Participants were required to rate their degree of concordance with each of the 30-item statements including, for example, “Manage the classroom during group activities with adequate feedback for all” (teaching), “Acting as a peer reviewer and complying with the deadlines established by the editorial team” (research), “Involve the external community in university extension activities” (extension), and “Perform administrative activities in parallel with teaching, research and extension activities” (management). A confirmatory factor analysis of the BLSES adopted here is presented as follows: χ²(394) = 1291.72, TLI = 0.91, CFI = 0.92, RMSEA = 0.052.

To evaluate burnout among participants we used the Maslach Burnout Inventory – Educators’ Survey¹ (MBI-ES) (Maslach and Jackson, 1986). This questionnaire is composed of 22 items arranged in three dimensions: emotional exhaustion (α = 0.92), depersonalization (α = 0.72) and personal accomplishment (α = 0.82). Items are rated using a 7-point Likert scale and include “I feel emotionally drained from my work” (emotional exhaustion), “I do not really care what happens to some students” (depersonalization), and “I have accomplished many worthwhile things in this job” (personal accomplishment). A confirmatory factor analysis of the MBI adopted here is presented as follows: χ²(195) = 1515.91, TLI = 0.91, CFI = 0.93, RMSEA = 0.063.

We also used the Brazilian version of World Health Organization Quality of Life Survey-Bref (WHOQOL-Bref) (Power and Kuyken, 1998; Fleck et al., 2000). This consists of 26 items rated using a 5-point Likert scale evaluating quality of life in four domains: physical (α = 0.84), psychological (α = 0.82), social relationships (α = 0.74), and environment (α = 0.77). Examples include “To what extent do you feel that physical pain prevents you from doing what you need to do?” (physical), “How much do you enjoy life?” (psychological), “How satisfied are you with your personal relationships?” (social), and “How satisfied are you with your transport?” (environment). A confirmatory factor analysis of the WHOQOL-Bref adopted here is presented as follows: χ²(236) = 1678.54, TLI = 0.90, CFI = 0.92, RMSEA = 0.060.

The validity and reliability of the questionnaires were evaluated conventionally using factor analysis after which the data collected was analyzed using both descriptive and inferential statistics (Field, 2013). We used SEM to explore the relationships between lecturer self-efficacy beliefs, quality of life and burnout and address the first research question (IBM AMOS 26.0; Neves, 2018). We conducted a series of exploratory analyses using Maximum Likelihood Estimation to test competing models and analyze different settings to determine the best fit adopting the following fit indices and preferred threshold values: χ²/df < 0.500, the comparative fit index (CFI) > 0.900, the Tucker–Lewis index (TLI) > 0.900, and the root mean square error of approximation (RMSEA) < 0.800 (Byrne, 2013).

To address the second research question and after the standardization of variables to z-scores (Everitt et al., 2011), we explored the data using cluster analysis (Ward’s method), a statistical technique that allows participants to be grouped based on their responses to the range of variables considered (IBM SPSS 25.0; Cohen, 1988; Field, 2013; Akoglu, 2018; Sharp et al., 2021). We also used chi-square to investigate associations between personal and professional variables and cluster membership (Akoglu, 2018).

We adopted SEM to investigate the relationships between lecturer self-efficacy, quality of life and burnout. We tested a working hypothesis that lecturer self-efficacy would predict quality of life and that quality of life would, in turn, predict burnout. Based on previous literature, we also hypothesized a direct relationship between lecturer self-efficacy and the different dimensions of burnout itself.

We chose to represent the constructs of lecturer self-efficacy and quality of life by single latent factors of a second order because they offer the advantage of constructing more parsimonious models to test (Parker et al., 2012). With regard to the burnout questionnaire (MBI-ES), however, it was decided to maintain its first-order factors since there are indications in the literature that these dimensions should not be combined (Skaalvik and Skaalvik, 2010). We also assumed that all MBI-ES factors would be correlated. We then conducted a series of exploratory analyses using Maximum Likelihood Estimation to test competing models to analyze different configurations to determine best fit. As Table 2 shows, and for the first model tested, we obtained fit indices of χ²(1519) = 4.697, TLI = 0.86, CFI = 0.87 and RMSEA = 0.047.

Following Byrne (2013), we then reviewed the model aimed at identifying high residual covariances between items and the theoretical relevance of specifying these covariances in the model. After modifying by specifying eight modification indices, we obtained the following values: χ²(1511) = 3.848, TLI = 0.894, CFI = 0.900 and RMSEA = 0.041. Next, we investigated regression values establishing as a criterion the removal of regression coefficients that were not significant for the model. After removing these, including a direct links between self-efficacy and the emotional exhaustion (r = 0.02) the depersonalization dimension of burnout (r = −0.09) dimension of burnout, the model presented the following final values: χ²(1513) = 3.851, TLI = 0.89, CFI = 0.90, RMSEA = 0.041. Full details of the final model are presented in Figure 1.

A chi-square test used to determine the differences between the first and the final models was significant (χ² = 16932.81; df = 6; p < 0.001).

Following Hayes (2009) we also bootstrapped the outcome to determine the significance of the direct and indirect mediation effects of quality of life models between self-efficacy and personal accomplishment, depersonalization and emotional exhaustion of the MBI-ES. The results indicate that all indirect effects were significant (Table 3).

Considering the complexity of involving three different questionnaires in arriving at a final model, we accept the model as a first exploration of the relationship between the constructs under analysis. The model demonstrates that, for this sample, lecturer self-efficacy is a positive predictor of quality of life (r = 0.46) which, in turn, negatively predicts burnout. Moreover, lecturer self-efficacy is also a predictor of the personal accomplishment factor of the MBI-ES scale (r = 0.26). Lecturer self-efficacy explains 22% of the variance in the participants’ perception of quality of life. The model also explains 53% of the variance in emotional exhaustion, 17% of the variance in depersonalization, and 23% of the variance in personal accomplishment.

We performed a cluster analysis using Ward’s method resulting in the identification of five main groups of participants. Further analysis was then performed at the level of individual factors which showed significant differences in all and with moderate to large effects: teaching F(4,1704) = 406.53, p < 0.001, η² = 0.49; research F(4,1704) = 193.62, p < 0.001, η² = 0.31; extension F(4,1704) = 195.17, p < 0.001, η² = 0.31; management F(4,1704) = 377.78, p < 0.001, η² = 0.47; emotional exhaustion F(4,1704) = 467.41, p < 0.001, η² = 0.52; depersonalization F(4,1704) = 100.91, p < 0.001, η² = 0.02; personal accomplishment F(4,1704) = 114.21, p < 0.001, η² = 0.21; the physical domain F(4,1704) = 358.32, p < 0.001, η² = 0.46; the psychological domain F(4,1704) = 376.57, p < 0.001, η² = 0.47; the social relationship domain F(4,1704) = 441.69, p < 0.001, η² = 0.51; and the environmental domain F(4,1704) = 226.47, p < 0.001, η² = 0.35. Although all groups differed slightly with regard to self-efficacy for teaching, research, extension and management activities, and in the social relationship domain of the WHOQOL-Bref, this difference was not significant, especially between Clusters 2 and 4 in the other dimensions. The cluster analysis is summarized as shown (Figure 2).

As observed in Figure 2, and as also hypothesized, Clusters 1, 2, and 5 associate higher scores of self-efficacy with lower scores of emotional exhaustion and depersonalization, and higher scores of personal accomplishment and all four factors of quality of life. In Clusters 3 and 4 some of the values are divergent, mainly in emotional exhaustion and depersonalization (MBI-ES) and social relationships (WHOQOL- Bref). Table 4 shows the means and standard deviations of each group in all factors of the three questionnaires.

Analyzing further (Table 5), Cluster 1, which we name here the Highly Efficacious/High Quality of Life and Accomplishment cluster, participants have high levels of self-efficacy and quality of life, with a high perception of personal accomplishment and low levels of exhaustion and depersonalization. With slightly more male than female participants (55.1%), these are participants who believe they cope well with the demands of the profession (76.5%) and feel more job satisfaction (96.6%). They are more active (76.1%), sleep well (75.2%), have few of the health symptoms listed when compared alongside the other clusters, and have a better perception of their own state of health (91.6% good or better).

Cluster 2, named here the Highly Efficacious/Moderate Quality of Life and Accomplishment cluster, is characterized by participants having high levels of self-efficacy with lower quality of life and personal accomplishment, and higher levels of exhaustion and depersonalization compared to Cluster 1. These were mostly male participants (71.2%) who reported a worsening sleep quality (only 56.2% claim sleeping well) and a greater number of adverse health-related symptoms (54.8% report anxiety), but still have a positive perception of their overall state of health (good or better 69.5%). Fewer participants believe they cope with the demands of the profession well (55.2%).

In Cluster 3, named the Moderately Efficacious/Exhausted cluster, participants presented average self-efficacy levels. Compared to Cluster 2, they show a large increase in emotional exhaustion and depersonalization in addition to loss of quality of life. Participants here, with well over half female (59.5%), perceived themselves as capable of performing the activities of teaching, research and extension effectively, but feel limited by physical fatigue and the demands arising from work. Only a small number reported actually coping well (25.6%). These participants present a reduction in the perception in the quality of their health (40.8% good or better), in satisfaction with work (66.6%), and present an increase in the number of adverse health-related symptoms in comparison with Clusters 1 and 2 (75.6% report anxiety).

Cluster 4, named the Low Efficacy/Low Social Relationships cluster, presents values close to those of Cluster 2 in several areas but differs in having lower self-efficacy across all four dimensions. Unlike Cluster 3, Cluster 4 is characterized by an improved perception of quality of life, with lower emotional exhaustion and depersonalization and poorer social relationships. Female and male participants are represented in almost equal numbers (50.2% male). Most perceive themselves to be healthy (68.9% good or better) and are satisfied with their profession (82.3%).

Cluster 5, named the Low Efficacy/Low Quality of Life cluster, presents participants with lower levels of self-efficacy in all four dimensions, as well as reduced personal accomplishment and perception of quality of life, combined with higher levels of exhaustion and depersonalization. This group is largely female in composition (60.4%), claiming not to have developed strategies to cope well with professional demands (only 13.1% think that they have). Most participants consider their general state of health as only satisfactory (66.4%), report performing fewer physical activities (56.8%) and exhibit poor quality of sleep (82.5%). They also have a greater number of adverse health-related symptoms compared to the other clusters (82.5% report anxiety). Very few also feel satisfied with the profession (13.1%).

Drawn from within Bandura’s Social Cognitive Theory, and the assumption that lecturer motivation and beliefs influence their well-being and contextual evaluation of the workplace, lecturer self-efficacy was indeed a positive predictor of quality of life, supporting the initial working hypothesis. Thus, lecturers who reported higher levels of self-efficacy perceived themselves with better quality of life. Our findings also indicate that lecturer self-efficacy is a direct positive predictor of personal accomplishment in burnout, a dimension that aggregates motivational factors, satisfaction and positive evaluation of work, corroborating the results of others (Briones et al., 2010; Bentea, 2017; Molero Jurado et al., 2018). Since higher levels of self-efficacy contribute to a positive mental state of engagement at work, self-efficacious individuals are therefore more likely to experience a greater number of successful work-related outcomes (Spontón et al., 2018). These findings are in accordance with those involving teachers in schools which have also found positive relationships between professional context, engagement and job satisfaction (Skaalvik and Skaalvik, 2010; Costa, 2012; Ventura et al., 2015; Chan et al., 2020; Matos et al., 2021). This raises the possibility of developing interventions that promote lecturer self-efficacy and, consequently, a more positive perception of quality of life thereby reducing the likelihood of suffering burnout (Fong et al., 2019).

Self-efficacy in the work environment can affect the way people perceive stress, anxiety and physical health. People with higher levels of self-efficacy, when exposed to unpleasant or exhausting situations, may experience less stress as they perceive themselves as more capable of dealing with the challenges imposed by the context (Bandura, 1997). As burnout is related to workplace demand, we anticipated a negative association between self-efficacy and burnout. However, we did not find the direct relationship witnessed elsewhere (Skaalvik and Skaalvik, 2010; Cao et al., 2018). Instead, we found a significant negative relationship between lecturer self-efficacy and emotional exhaustion and depersonalization mediated by quality of life. Thus, quality of life impacts negatively on burnout syndrome while positively affected by self-efficacy. Bandura (1997, p. 464) argues, however, that “occupational stress is not just an employee problem. Certain organizational conditions can undermine employees’ beliefs in their occupational capabilities.” Higher education working conditions are also important, pointing to the university environment as highly exhausting and demanding. This needs to be discussed and addressed and will be further explored in the next sections (Bandura, 1997; Alves, 2017; Cardoso Júnior et al., 2018; Alves et al., 2019; Hall et al., 2019).

Recent studies have related teacher self-efficacy with their mental health, with implications on their behavior and attitudes toward work (Von Muenchhausen et al., 2021). Therefore, lecturer self-efficacy could be understood as a valuable personal resource that influences on the adoption of positive coping skills, proactive behavior and positive emotions as whole. In this sense, researchers have demonstrated that it is possible to develop training processes for lecturers with the purpose of increasing their self-efficacy beliefs through the use of the sources of information (Postareff et al., 2007; Fong et al., 2019). Thus, the main importance of these results may lie in developing intervention processes that promote lecturer self-efficacy, with a positive perception of quality of life, which could reduce the possibility of suffering with burnout. In addition to findings of studies that point the importance of lecturer self-efficacy beliefs on student learning, the results show that these beliefs are pivotal for lecturers themselves (Matos et al., 2021).

In addition to developing strategies for increasing self-efficacy, it is essential that universities also develop strategies to help improve the work environment, dealing with excessive demands that act as the precursors of emotional exhaustion (Hall et al., 2019). Considering the international situation facing universities post-pandemic, and the changes required to build “new normals,” it might be suggested that addressing self-efficacy and collective efficacy provides at least one new pathway over the coming years.

Cluster analysis contributed to understanding variations in self-efficacy beliefs among participants, with participants brought together into five groups characterized by decreasing values of self-efficacy across each of its four dimensions. These differences were not straightforward when also considering burnout and quality of life together.