The job and life satisfaction of teachers: a social cognitive model integrating teachers’ burnout, self-efficacy, dispositional optimism, and social support

The purpose of the study was to test a model of factors predicting teachers’ job and life satisfaction, burnout, dispositional optimism, social support, perceived workload, and self-efficacy. The model extends Lent and Brown’s (J Voc Behav 69(2):236–247, https://doi.org/10.1016/j.jvb.2006.02.006, 2006; J Career Assess 16(1):6–21, https://doi.org/10.1177/1069072707305769, 2008) social cognitive model of the interaction of sources of job and life satisfaction. Specifically, burnout, a condition with a high incidence rate among teachers, was included. The participants were 676 Swiss teachers. Structural equation modeling was used to analyze the data. The results revealed the differential effect of the variables considered on teachers' burnout and job satisfaction, as well as their life satisfaction. Dispositional optimism, social support, and perceived workload might reduce the risk of teacher burnout; dispositional optimism, social support, and teacher self-efficacy seem to positively affect job satisfaction; and dispositional optimism alone, together with burnout and job satisfaction, directly relates to teachers’ life satisfaction. Practical implications of these results are discussed.


Introduction
Teachers' job and life satisfaction (Caprara et al., 2006;Lent et al., 2011), as well as teachers' stress and burnout (Schwarzer & Hallum, 2008;Skaalvik & Skaalvik, 2016), have been the focus of a growing number of studies in recent years. Brown (2006, 2008) developed a multifactorial model that explains job satisfaction starting from their social cognitive career theory (Lent et al., 1994). This social cognitive model of job satisfaction (SCM-JS) integrates personality/affective traits, participation in/progress at goal-directed activities, work-related self-efficacy, perceived work conditions and outcome expectations, and goal-and efficacy-relevant environmental supports or obstacles and clarifies how these factors relate and contribute to explaining job satisfaction and life satisfaction. In particular, according to Brown's (2006, 2008) SCM-JS, job satisfaction is influenced by goal-directed activity, self-efficacy, work conditions and outcomes, goal-and efficacy-relevant environmental supports and obstacles, and personality and affective traits. Job satisfaction is also assumed to impact global life satisfaction.
In the past, a large number of studies have been devoted to the study of teachers' burnout and its association with internalizing and externalizing behaviors (Maslach et al., 2001;Melamed et al., 2006). While teachers' job satisfaction can be seen as a positive indicator of teachers' well-being, burnout is undoubtedly a negative indicator of teachers' well-being. Multifactorial studies on teachers' well-being, which consider different personal and contextual variables as predictors of both burnout and job and life satisfaction, are lacking. Hence, our aim was to describe the relation between teachers' dispositional optimism, perception of social support, self-efficacy, perceived workload, burnout, and job and life satisfaction by testing a model based on the Brown (2006, 2008) SCM-JS. We also wanted to verify that the model was invariant between female and male teachers.

Life and job satisfaction
According to Diener (1984), life satisfaction is the judgment that individuals give about their well-being and quality of life based on personal criteria. A large body of studies has demonstrated that life satisfaction is associated with different positive personal, psychological, and social outcomes (DeNeve & Cooper, 1998;Diener & Tay, 2012;Pavot & Diener, 1993). Some longitudinal studies have suggested that life satisfaction not only is influenced by but also influences these positive outcomes. Indeed, higher life satisfaction seems to predict higher income (De Neve & Oswald, 1 3 2012), greater job success (Boehm & Lyubomirsky, 2008;Luhmann et al., 2013), better relationships (Oishi et al., 2007), a higher likelihood of marriage and childbirth and a lower likelihood of marital separation and relocation (Luhmann et al., 2013), fewer health problems, and greater longevity (Diener & Chan, 2011). Even if the processes underlying these associations are still unclear, studies seem to indicate that life satisfaction drives people toward the adoption of personal goals and hence a higher level of control over their life circumstances (Luhmann & Hennecke, 2017). Thus, life satisfaction is an important indicator of present well-being and might be seen as a resource for future well-being.
Job satisfaction is an important factor to consider when studying workers' desirable outcomes and well-being. Job satisfaction is defined as "a pleasurable or positive emotional state resulting from the appraisal of one's job or job experiences" (Locke, 1976(Locke, , p. 1300 or, more briefly, "the extent to which people like (satisfaction) or dislike (dissatisfaction) their jobs" (Spector, 1997, p. 2). Studies have indicated that job satisfaction is influenced by characteristics of the work environment (Fairbrother & Warn, 2003) and individual dispositions such as personality traits (Steel et al., 2019). Specifically, with regard to teachers, self-regulation processes such as perceived self-efficacy (Caprara et al., 2006) have been found to predict job satisfaction, which in turn influences teachers' retention (Ingersoll, 2001) and relations to students (Veldman et al., 2013).
The only study that tested the whole SCM-JS on teachers found that life satisfaction is predicted by positive affect and goal-directed activity, and job satisfaction is predicted by positive affect, efficacy support, and work conditions but is not related to self-efficacy or goal-directed activity (Lent et al., 2011). In this study, job satisfaction influenced life satisfaction. However, as also reported by Lent and Brown (2008), studies suggest that the link between these two variables might be bidirectional (Schmitt & Pulakos, 1985) and that the path from life satisfaction to job satisfaction may be the strongest (Bialowolski & Weziak-Bialowolska, 2020;Bowling et al., 2010;Unanue et al., 2017).

Burnout
Burnout can be defined as a state of exhaustion, i.e., physical, emotional, and mental (Kristensen et al., 2005), which derives from being exposed to a strongly demanding work environment (Schaufeli & Greenglass, 2001, p. 501). Burnout occurs only in the work domain and is a result of chronic stress (Milfont et al., 2008). Given its association with several negative conditions and behaviors, such as depression, anxiety, substance abuse, decreased performance, and increased health problems (Maslach et al., 2001;Melamed et al., 2006), burnout has been recognized to be a serious problem for individuals, organizations, and society (Schaufeli et al., 2009). Among teachers, compared to other professional categories, several studies have found higher levels of emotional exhaustion (Schaufeli & Enzmann, 1998), as well as a higher incidence rate of psychiatric disorders (Lodolo D'Oria et al., 2004). Previous studies have provided several examples of factors related to teacher burnout, such as feelings of isolation and lack of support from colleagues, perceived work overload, role ambiguity or role conflict, lack of autonomy, and problems with classroom management and discipline (Alarcon, 2011;Avanzi et al., 2018;Grayson & Alvarez, 2008;Yildrim, 2008). Some studies have also demonstrated a negative association between life satisfaction and burnout (David & Quintao, 2012;Hayes & Weathington, 2007). Demerouti et al. (2000) showed that burnout has a mediating effect between job demands or resources and life satisfaction. Hakanen and Schaufeli (2012) specified that burnout could predict depressive symptoms and life dissatisfaction but not vice versa. Evidence of a relation with job satisfaction has also been found in salespersons (Rutherford et al., 2011;Singh et al., 1994), while few studies have been conducted among teachers (Skaalvik & Skaalvik, 2009). Finally, previous studies have highlighted that social support might serve as a coping strategy to adapt to a stressful working environment (Halbesleben, 2006;Viswesvaran et al., 1999) and a useful resource to contrast a condition of perceived workload and burnout (Maslach et al., 2001).

Dispositional optimism
Dispositional optimism is the expectation that one's own outcomes will generally be positive. Conversely, pessimism is the expectation of negative outcomes. This cognitive construct influences motivation (Carver & Scheier, 2014), whereas optimistic people have the tendency to exert effort, pessimistic people disengage from it. Previous literature has confirmed that optimists are more successful in work environments (Seligman & Schulman, 1986), manage difficult situations more effectively (Wrosch & Scheier, 2003), better balance effort expenditures (Segerstrom, 2006), have better social relationships (Neff & Geers, 2013;Srivastava et al., 2006) and health (Kim et al., 2011), and are less depressed (Schweizer, et al., 1999) than pessimists. Indeed, dispositional optimism has been found to be positively associated with life satisfaction (Harju & Bolen, 1998;Malinauskas & Vaicekouskas, 2013;Reed, 2016), self-efficacy and perceived social support (Karademas, 2006) and negatively associated with perceived workload (Gardner & Parkinson, 2011;Reed, 2016) and burnout (Hayes & Weathington, 2007). These findings are consistent with the SCM-JS, which assumes that personality and affective traits are linked to job and life satisfaction both directly and indirectly through several cognitive, affective, behavioral, and social paths.

Social support
Social support is "the availability of helping relationships and the quality of those relationships" (Leavy, 1983, p. 5). Johnson and Hall (1988) thought social support to be highly important to explain job strain; hence, they expanded the job demandcontrol model of job strain (Karasek, 1979) to integrate this factor. According to the literature, social support may modify the impact of psychological demands, both in (Viswesvaran et al., 1999) and out of the work context (Cobb, 1976). Indeed, research has connected perceived social support negatively with burnout (Prins 1 3 et al., 2007) and positively with measures of subjective well-being (Siedlecki et al., 2013) and optimism (Karademas, 2006). Studies conducted on teachers have shown that interpersonal relationships play a key role in teachers' work (Van Droogenbroeck et al., 2014) and that positive and satisfying relationships with colleagues, parents, and students moderate burnout (Cano-García et al., 2005;Doudin et al., 2011;Gavish & Friedman, 2010). Specifically, in studies based on the three-dimensional conceptualization of burnout, which includes emotional exhaustion, depersonalization, and reduced personal accomplishment (Maslach & Jackson, 1986), work-related sources of support, i.e., those coming from professional figures such as colleagues and superiors, because they directly affect job demand perception, strongly relate to the exhaustion dimension of burnout (Betoret, 2006;Halbesleben, 2006). Coherently with these results, in the SCM-JS, social support directly influences self-efficacy, work condition perceptions, and job satisfaction (and burnout) and is influenced by personality traits.

Perceived workload
According to the models by Karasek (1979Karasek ( , 1998job demand-control) and Demerouti et al., (2001;job demands-resources), which define job strain as a consequence of the discrepancy between job demands and the autonomy or control over the job or the resources that are available to individuals to face those demands, respectively, perceived workload might be a relevant factor that influences well-being. There are indeed many studies confirming an association between workload and, respectively, emotional exhaustion and burnout (Alarcon, 2011;Kokkinos, 2007). Instead, unlike links theorized in SCM-JS, the results of a regression analysis performed in a recent meta-analysis investigating factors associated with and consequences of workload, led the authors to conclude that workload did not have a relation with job satisfaction after controlling for the effects of role ambiguity and role overload (Bowling et al., 2015). In the same meta-analytic study, the authors found that workload was negatively associated with social support, both from supervisors and coworkers, and positively associated with the negative affectivity trait. In the SCM-JS, perceived workload falls into the work conditions category. Hence, it should be predicted by social support and self-efficacy and should predict job satisfaction and burnout. Although few studies have suggested that perceived workload can directly influence life satisfaction (Goh et al., 2015), according to SCM-JS, this influence occurs indirectly through job satisfaction.

Teachers' self-efficacy
According to Bandura's theory regarding the sources of self-efficacy (1997), teachers make judgments about their self-efficacy, i.e., about their own ability and skills in teaching even in unfavorable conditions, based on the perceptions of their past teaching experiences, the failures or successes of their fellow teachers that serve as vicarious experiences, the verbal incentive present in their work environment, such as that from colleagues and principals, the physiological state related to the emotions and the level of excitement experienced in teaching practice. Thus, together with personal mastery experience, social support, and good relationships in the working context are crucial to enhancing self-efficacy (Korte & Simonsen, 2018;Usher & Pajares, 2008). Additionally, according to Bandura (2006), self-efficacy determines how environmental opportunities and impediments are perceived. Moreover, selfefficacy influences individuals' goals and behaviors. In particular, efficacy beliefs influence the choice of activities, the effort expended on an activity, and how long the individual will persevere when facing obstacles. Consequently, people with high levels of self-efficacy beliefs are expected to put themselves in more challenging situations, which may also lead to higher success in job and life. Indeed, teachers' self-efficacy has been linked with an increased persistence in working with challenging students and has been shown to affect teachers' commitment, enthusiasm, practices, and teaching behaviors (Skaalvik & Skaalvik, 2007). Furthermore, concerning its relation with well-being variables, research has shown that self-efficacy is related to teachers' job satisfaction, stress, and burnout (Caprara et al., 2006;Klassen & Chiu, 2010;Malinen & Savolainen, 2016;Schwarzer & Hallum, 2008;Skaalvik & Skaalvik, 2016).
Consistent with the above, in the SCM-JS, self-efficacy is influenced by support and personality traits/positive affectivity and directly and positively impacts goal progress, work conditions, and job satisfaction; its effect on life satisfaction goes through job satisfaction (and burnout). Whereas the study of Duffy and Lent (2009) on US teachers confirmed the direct relation between teacher self-efficacy and job satisfaction, the study of Lent et al. (2011) on Italian teachers did not find a direct link between these variables but only an indirect link through work conditions.

Current study
As previously mentioned, only a few studies have tested all or most of the relationships theorized in Brown's (2006, 2008) model on teachers (Duffy & Lent, 2009;Lent et al., 2011), and the results of these studies are not always consistent. From the analysis of the literature just presented, a few questions emerge for which we would like to provide some additional explanatory information with this study. First, the direction of the relation between job and life satisfaction is not clear; the findings are indeed inconsistent (e.g., Bialowolski & Weziak-Bialowolska, 2020;Bowling et al., 2010;Schmitt & Pulakos, 1985;Unanue et al., 2017). Hence, in this study, both links are tested alternatively. Second, it is unclear whether the construct of burnout specifically relates to job satisfaction and whether burnout and life satisfaction are also linked directly (David & Quintao, 2012;Hakanen & Schaufeli, 2012;Hayes & Weathington, 2007) or only indirectly through job satisfaction. Thus, in this study, these associations are explored. Third, according to literature (Gardner & Parkinson, 2011;Reed, 2016), unlike what appears in SCM-JS, a direct relationship between optimism and perceived workload can be hypothesized. Hence, this link is tested in this study. Fourth, the relationship between perceived workload and life satisfaction is not clear whether it is direct (Goh et al., 2015) or mediated by job 1 3 satisfaction, as stated in the Brown (2006, 2008) model, but not supported by Bowling et al. (2015). Finally, although partially related to the fourth research gap, the link between teacher self-efficacy and job satisfaction is also unclear, particularly whether it is direct (Duffy & Lent, 2009) or partially/fully mediated by perceived workload (work conditions) (Lent et al., 2011). Hence, these mediation pathways are also tested in this study.
In the model that we wanted to test with this study, burnout was considered an indicator of occupational well-being as job satisfaction; thus, it was set as directly predicted by self-efficacy, support from colleagues, dispositional optimism, and workload. With respect to the other relationships, we adhered to the Brown (2006, 2008) SCM-JS. The links hypothesized are summarized in Figure 1.

Participants and procedure
The study was conducted in the Swiss state of the canton of Ticino, in the southern region of Switzerland. Teachers working in Ticino undergo professional training that lasts for 2 or 3 years, which is equivalent to a bachelor's or master's degree, depending on the school sector for which they are qualified. For the working environment, primary, lower secondary, upper secondary, and vocational schoolteachers generally work in school institutes with a school principal, administration staff, and several colleagues, while preprimary schoolteachers generally work in small school sites with only a few colleagues and report to a school principal often working elsewhere.
The participants were teachers from 24 public school institutes of primary, secondary, and vocational schools, conveniently selected over the Ticino territory and Figure 1 A model to predict teacher well-being based on Brown (2006, 2008) model. Note Dotted line = links added to Lent and Brown model; + hypothesis of a positive relationship; − hypothesis of a negative relationship identified on the basis of school sector, school size, and location criteria. Teachers were contacted through e-mail invitations, where information about the study and the guarantee of anonymity of the study were provided. Anonymous data were collected through an online questionnaire. To have access to the questionnaire, an anonymous link was sent to participants via e-mail invitations.
The data collection involved 676 teachers, which was equal to 8% of the total population of active teachers in Ticino; 375 were females (55.5%), and 301 were men (44.5%). Most of them (85.8%) had a percentage of working time equal to or above 50% of a full-time contract. The mean age was 44.53 years (SD = 10.27), and the mean school tenure was 15.7 years (SD = 11.49). Teachers participating in the study worked in preprimary and primary schools (18%), lower secondary schools (18.3%), upper secondary schools (29.3%), and vocational schools (34.3%). For the above variables, the proportions were similar to those of the reference population.

Life satisfaction
Global life satisfaction was measured with the Italian version of the Satisfaction with Life Scale (Di Fabio & Palazzeschi, 2012;Diener et al., 1985). The scale consists of 5 items rated on a 7-point Likert-type scale, with responses ranging from 1 (strongly disagree) to 7 (strongly agree). An example item is "In most ways my life is close to my ideal". The Cronbach's alpha was 0.87 for the original English scale (Diener et al., 1985) and 0.88 for the Italian scale (Di Fabio & Palazzeschi, 2012).

Job satisfaction
To assess job satisfaction, 6 items were created ad hoc by educational psychologists based on previous job satisfaction scale items. The items are rated using a 4-point Likert-type scale, with responses ranging from 1 ("Don't agree at all") to 4 ("Totally agree"). Two item examples are "The work I do meets my expectations" and "I am generally satisfied with my profession".

Burnout
To measure burnout, we used the 5-item version of the work-burnout subscale of the Copenhagen Burnout Inventory (Kristensen et al., 2005) developed and validated in Italian by Marcionetti et al. (2018). The items are rated using a 5-point Likert-type scale, with responses ranging from 1 ("never, almost never"/ "to a low degree, to a very low degree") to 5 ("always"/ "to a very high degree"). An example item on the work-burnout scale is "Is your work emotionally exhausting?". The Cronbach's alpha for the original English subscale was 0.87 for the work-burnout scale. The omega value (McDonald, 1999) of the Italian short version of the scale was 0.84.

Optimism
Dispositional optimism was measured using the Italian version of the Revised Life Orientation Test (Giannini et al., 2008;Scheier et al., 1994). The scale consists of 6 items (plus 4 filler items) rated on a 5-point Likert-type scale, with responses ranging from 1 ("strongly disagree") to 5 ("strongly agree"). An example item is "I'm always optimistic about my future". The Cronbach's alpha of the original English scale was 0.78, and 0.81 for the Italian translation.

Social support
To measure perceived support from colleagues, the 4-item scale of Edwards et al. (2008), translated and validated in Italian by Toderi et al. (2013), was used. The items are rated on a 5-point Likert-type scale, with responses ranging from 1 ("never") to 5 ("always"). An example item is "I get help and support I need from colleagues". The Cronbach's alpha was 0.82 for the original English scale and 0.89 for the Italian scale.

Workload
Workload was measured with the Italian short version of the workload scale (Edwards et al., 2008;Toderi et al., 2013). We discarded two of the eight items originally included in the scale because of a lack of fit in the local school context. Items are rated on a 5-point Likert-type scale, with responses ranging from 1 ("never") to 5 ("always"). An item example is "I am pressured to work long hours". The Cronbach's alpha was 0.87 for the original English scale and 0.84 for the Italian scale.

Teacher self-efficacy
To assess teacher self-efficacy, the Italian version of the Teacher Self-Efficacy Scale (Marcionetti & Castelli, in press) originally developed by Schwarzer et al. (1999) was used. The scale includes ten items rated on a 4-point Likert scale from 1 ("not true at all") to 4 ("totally true"). An example item is "If I try hard enough, I know that I can exert a positive influence on both the personal and academic development of my students". The Cronbach's alpha was between 0.76 and 0.82 for the original scale (validated on three samples of teachers) and 0.86 for the Italian scale.

Data analysis
First, confirmatory factor analysis (CFA) was performed for each scale. A covariance matrix was generated to run the models. To assess the adequateness of measurement models with small degrees of freedom (Kenny et al., 2015), we relied on the comparative fit index (CFI) and the Tucker-Lewis index (TLI), which indicate good fit when values are approximately 0.90 or above (Medsker et al., 1994). We also inspected factor loadings that were adequate when above 0.40. Internal consistency was assessed with Cronbach's alpha. Second, means, standard deviations, and correlations between the variables were calculated. Then, to assess the hypotheses illustrated in Figure 1, we started by specifying a structural equation model (SEM) for the full sample. The original scale items were used as observed variables, and the covariance matrix generated to run the CFA was used to run SEM models. To assess model fit, we used the χ 2 per degrees of freedom (χ 2 /df), which indicates a good fit when the value is equal to or below 3 (Kline, 2005), the already mentioned CFI and TLI, and the root mean square error of approximation (RMSEA) value, which is an indicator of good fit when it is below 0.05, although a RMSEA value of approximately 0.08 or less is also acceptable (Byrne, 2010). Second, a multiplegroup model was tested to investigate whether the estimated effects were similar for female and male teachers. To do this, measurement invariance was tested by inspecting changes in fit across the configural, weak invariance, and structural models. As recommended by Chen (2007), the assumption of invariance across models was considered to be tenable if ΔCFI < 0.01 and ΔRMSEA < 0.015.

Results
CFA with acceptable/good TLI and CFI values and factor loadings well above 0.40 confirmed the adequateness of the measures used. Particular attention was given to the ad hoc created job satisfaction scale, on which the factor loadings ranged from 0.51 to 0.84, and the TLI and CFI were 0.91 and 0.95, respectively, and to the adapted workload scale, on which the factor loadings ranged from 0.59 to 0.79, and the TLI and CFI were 0.94 and 0.97, respectively. The Cronbach's alpha, means, standard deviations, and correlations between the variables included in this study are presented in Table 1. The Cronbach's alpha ranged from 0.80 to 0.88, which indicated the good reliability of the scales used. Concerning correlations, the lowest were between self-efficacy and perceived workload (r = 0.18) and between social support and self-efficacy (r = 0.19), and the highest were between burnout and perceived workload and between job satisfaction and self-efficacy (both r = 0.55).
SEM models were then specified, including all constructs as latent variables. The first model (Model 1), including all relations exposed in Figure 1, reached a TLI value slightly under acceptability, χ 2 (800) = 1987.572, p < 0.000, TLI = 0.893, CFI = 0.900, and RMSEA = 0.048. Hence, following modification indices, we finally ran a second model (Model 2) where items 1 and 3 and items 9 and 10 of the selfefficacy scale, items 1 and 4 of the optimism scale, and items 3 and 4 of the support scale correlated. The fit of this model was adequate, χ 2 (796) = 1686.909, p < 0.000, TLI = 0.919, CFI = 0.925, and RMSEA = 0.042. Estimates for significant and nonsignificant pathways are reported in Table 2. Some paths were nonsignificant; hence, a third model (Model 3) was run where nonsignificant relations were removed. The fit of this model was also good, χ 2 (801) = 1692.753, p < 0.000, TLI = 0.920, CFI = 0.925, and RMSEA = 0.042. A fourth model, which had the same indices of fit as the last one, was specified with life satisfaction predicting job satisfaction to also test this relation. The beta value of Table 1 Cronbach's alpha, mean, standard deviation, and correlation matrix On this model, using multiple-group SEM, we tested gender invariance to confirm that the model was adequate for both female and male teachers. First, weak factorial invariance (all factor loadings constrained to be equal between the two groups) was analyzed, testing for differences across the weak and configural models. The values obtained for the configural model [χ 2 (1602) = 2645.879, p < 0.000, TLI = 0.907, CFI = 0.914, and RMSEA = 0.032] were very similar to those observed for the weak invariance model [χ 2 (1637) = 2679.945, p < 0.000, TLI = 0.910, CFI = 0.914, and RMSEA = 0.032]. These analyses thus confirmed a weak factor invariance (ΔCFI and ΔRMSEA = 0.000). Second, a model with pathways constrained to be equal across the two groups was tested to investigate whether there were any differences ascribable to pathways between latent variables. The model constraining pathways to be equal was not significantly worse than the weak invariance model [χ 2 (1693) = 2824.928, p < 0.000, TLI = 0.905, CFI = 0.907, RMSEA = 0.032; In the third model (Model 3), whose standardized beta values are presented in Figure 2, dispositional optimism directly and positively influences the perception of being supported by colleagues, teacher self-efficacy, and life satisfaction and negatively influences burnout and the perception of workload; the perception of support from colleagues directly and positively influences teacher self-efficacy and job satisfaction and negatively influences the perception of workload; and workload only but highly affects burnout. Teachers' self-efficacy directly and positively influences job satisfaction and, negatively, burnout. Burnout has a negative effect on job satisfaction and life satisfaction. Finally, job satisfaction has a positive effect on life satisfaction. The model permitted us to explain 52%, 48%, and 49% of burnout, job satisfaction, and life satisfaction variance, respectively.

Discussion
In this study, we tested a model based on Brown's (2006, 2008) model, which describes the relation between teachers' dispositional optimism, perception of social support, self-efficacy, workload, burnout, job and life satisfaction, and gender invariance. In the model, the construct of burnout was included as another important indicator of occupational well-being for teachers, in addition to job satisfaction. The aim of the study was to provide additional information about relationships theorized in the Brown (2006, 2008) SCM-JS (see Figure 1) and, in particular, on those links that were only partially supported by empirical studies (see "Current study" section). The results are discussed by considering all the hypothesized relationships.

Job satisfaction and life satisfaction
In line with SCM-JS and with our hypothesis (Figure 1), in our model life satisfaction is directly influenced by job satisfaction. However, in line with previous literature, the inverted link between life satisfaction and job satisfaction has also emerged as promising (Bialowolski & Weziak-Bialowolska, 2020;Bowling et al., 2010;Schmitt & Pulakos, 1985;Unanue et al., 2017). This finding further corroborates the hypothesis that these variables might positively influence one another. Thus, if good job satisfaction allows people to be happier with their lives in general, life satisfaction, in turn, could affect their job satisfaction by allowing people to adopt personal goals (Luhmann & Hennecke, 2017).

Burnout, job satisfaction, and life satisfaction
Our model suggests that burnout relates directly to both job satisfaction (Rutherford et al., 2011;Singh et al., 1994) and life satisfaction (David & Quintao, 2012;Hakanen & Schaufeli, 2012;Hayes & Weathington, 2007). To our knowledge, the relation of burnout with job satisfaction has rarely been assessed in teachers. This result also suggests that job satisfaction might be a partial mediator of the relationship between burnout and life satisfaction. Hence, burnout might be important to explain low levels of job (and life) satisfaction in teachers and, given its high incidence in this professional category (Schaufeli & Enzmann, 1998), should be included in models explaining the processes underlying teachers' well-being.

The effect of workload on burnout, job satisfaction, and life satisfaction
In contrast to the results of the study by Goh et al. (2015), in our model the relationship between teachers' perceived workload and life satisfaction appears to be only indirect, as stated in the SCM-JS. However, the effect is mediated by burnout and not by job satisfaction, which is in line with the results of Bowling et al. (2015), which show that workload does not have a direct association with job satisfaction.

Self-efficacy, workload, and job satisfaction
In contrast to the only other study considering the Lent and Brown (2008) model's factors to study job satisfaction in teachers (Lent et al., 2011), our study strongly confirms (with a β value of 0.50) the direct relation between self-efficacy and job satisfaction. This means that among Swiss teachers, the perception of being good at their job directly increases professional satisfaction. However, unlike Brown's (2006, 2008) model and the results of Lent et al. (2011), in our model teachers' self-efficacy does not directly influence the perception of workload. Hence, Swiss teachers' efficacy beliefs can surely influence their choice of activities, the effort they expend on an activity, and how long they will persevere when facing obstacles (Bandura, 2006); however, those beliefs are less likely to affect their perception of workload.

3
The effect of perceived support on self-efficacy, workload, and job satisfaction Moreover, in our model, it emerged that both teachers' self-efficacy and workload perception, as well as job satisfaction, were directly influenced by the perception of receiving support from colleagues. This result is in line with the SCM-JS. This finding is also in line with the results of previous studies, which have highlighted that positive interpersonal relationships play a key role in teachers' work and contribute to increasing teachers' self-efficacy (Beasley et al., 2013;Huang et al., 2019;Van Droogenbroeck et al., 2014), reducing the work strain experienced, and mitigating perceived stressors (Bowling et al., 2015;Cano-García et al., 2005;Gavish & Friedman, 2010;Viswesvaran et al., 1999).

The effect of dispositional optimism on social support, self-efficacy, workload, burnout, and life satisfaction
It is also interesting to observe the effect of personality traits that emerged in our model that are mainly in line with SCM-JS. Consistently with previous studies, perceived support from colleagues and self-efficacy (Karademas, 2006) are positively affected by dispositional optimism. In line with the literature (Gardner & Parkinson, 2011;Reed, 2016), our study also suggests that a direct relationship between dispositional optimism (personality) and perceived workload (work conditions) can be added to SCM-JS. Indeed, it appears that more optimistic teachers are likely to perceive less workload. Dispositional optimism, according to the Hayes and Weathington (2007) study, also affects burnout and in our model is the only direct predictor of life satisfaction (Harju & Bolen, 1998;Malinauskas & Vaicekouskas, 2013;Reed, 2016), aside from the two other well-being indicators, i.e., burnout and job satisfaction. This personality characteristic thus seems to be an important antecedent, both direct and indirect, of (teacher) well-being. Indeed, in line with SCM-JS, it seems to increase the chance to benefit from colleagues' support, to have higher feelings of self-efficacy, and to increase levels of life satisfaction. Even if a direct relation between dispositional optimism and job satisfaction did not emerge, this link might be mediated by perceptions of colleagues' support, self-efficacy, and burnout. Thus, the results of our study confirm that even among teachers, optimists, compared to pessimists, have better social relationships (Neff & Geers, 2013;Srivastava et al., 2006), can handle difficult situations better (Wrosch & Scheier, 2003), and have better mental and physical health (Kim et al., 2011;Schweizer et al., 1999), which increases their likelihood of success at work (Seligman & Schulman, 1986).

Antecedents of burnout and job satisfaction
As previously stated, unlike with the SCM-JS, in our study, dispositional optimism does not affect job satisfaction, whereas in line with previous studies it does affect burnout (Hayes & Weathington, 2007). Similarly, in our model and in line with Bowling et al. (2015), workload does not directly affect job satisfaction, whereas in line with the studies from Alarcon (2011) and Avanzi et al. (2018), it does affect burnout. In the same perspective, unlike other empirical findings (Betoret, 2006;Cano-García et al., 2005;Doudin et al., 2011;Gavish & Friedman, 2010;Halbesleben, 2006;Prins et al., 2007), in our model, perceived support from colleagues does not have a direct effect on burnout, whereas in line with the SCM-JS, it does affect job satisfaction. Hence, the only construct that, according to the SCM-JS and previous literature (Caprara et al., 2006;Klassen & Chiu, 2010;Malinen & Savolainen, 2016;Schwarzer & Hallum, 2008;Skaalvik & Skaalvik, 2016), seems to affect both burnout and job satisfaction is self-efficacy. The link is, however, stronger with job satisfaction (β = 0.50) than with burnout (β = − 0.14). Thus, adding the construct of burnout into the model allowed us to clarify that burnout is mostly affected by perceived working conditions (workload) and personality, job satisfaction is mostly influenced by self-efficacy beliefs. Our model thus allowed us to further clarify both the relationship between burnout and job and life satisfaction and the potential direct antecedents of burnout and job and life satisfaction.

Limitations of study and suggestions for future research
This study of teacher well-being, which builds on Brown's (2006, 2008) SCM-JS and adds the important construct of burnout, allows for clarification of how teachers' dispositional optimism, perceptions of social support, self-efficacy, perceived workload, burnout, and job and life satisfaction connect. However, the crosssectional design of this study does not permit us to make causality inferences or to be sure of the mediation effects that emerged. This is the strongest limitation of this research, which further highlights the need for longitudinal studies that investigate more in depth the mediation effects that emerged, as well as the life satisfactionjob satisfaction relationship where the temporal dimension is taken into account. In addition, a larger sample would have allowed further analysis, such as the model invariance test between school sectors. The differences that emerged between some studies conducted on teachers could in fact be due to differences related to the school sector in which teachers work (and therefore strongly dependent on the composition of the sample). For example, differences in the support available in school between primary school teachers who work in small school institutes and other teachers who work in larger schools with more colleagues might lead to slightly different results for this group of teachers. Finally, this model should be further tested with teachers working in different educational contexts to ensure that it can be generalized to contexts outside Switzerland. However, the consistency of much of the relationships that emerged in our model with those hypothesized by Brown (2006, 2008) and with those confirmed in the existing literature allows us to infer good reliability for the results of our study.

Conclusion
Regarding the practical implications, the results of this study indicate that fostering teacher optimism, self-efficacy, and good relationships with colleagues might be crucial to reduce the risk of burnout and increase teacher job (and life) satisfaction. Even if trying to modify a personality characteristic such as dispositional optimism seems to be difficult, studies have shown that it is, however, possible to learn to be more optimistic (Seligman, 1998). The perceived school climate has been related to teachers' job satisfaction; this relationship is partly mediated by self-efficacy (Malinen & Savolainen, 2016). Creating a positive climate at school might also foster an optimistic view on (work and) life. A positive climate where teachers have good relationships with their colleagues, which also means the possibility to discuss and reflect together about their daily work matters and find solutions to problematic situations, thereby increasing their self-efficacy, might be important to diminishing the risk of burnout and increasing teacher job and life satisfaction.