Received: 3 August 2016 Revised: 29 September 2016 Accepted: 12 November 2016

DOI: 10.1002/smi.2737

RESEARCH ARTICLE

Burnout and depressive symptoms in teachers: Factor structure

and construct validity of the Maslach Burnout inventory‐
educators survey among elementary and secondary school
teachers in Hungary
Réka Szigeti1,2 | Noémi Balázs1,2 | Réka Bikfalvi2 | Róbert Urbán3

1
Doctoral School of Psychology, Eötvös
Loránd University (ELTE PPK), Budapest, Abstract
Hungary This study validated the Hungarian version of the Maslach Burnout Inventory–Educators
2
Foundation of Mental Health for Wellbeing, Survey on a sample of n = 211 elementary and secondary teachers. To test factorial validity,
Budapest, Hungary we ran a series of confirmatory analysis with eight models. The best fitting model was the
3
Institute of Psychology, Eötvös Loránd bifactor model with general burnout and three specific factors: emotional exhaustion, deper-
University (ELTE PPK), Budapest, Hungary
sonalization, and personal accomplishment. Analyzing the covariates revealed that gender and
Correspondence
age were not associated with burnout, but depressive symptoms and overcommitment had a
Réka Szigeti, Doctoral School of Psychology,
Institute of Psychology, Eötvös Loránd significant relationship with general burnout, and overcommitment was related to emotional
University (ELTE PPK), Izabella u. 46, Budapest exhaustion as well.
1064, Hungary.
Email: [email protected]
KEY W ORDS

burnout, depression, job stress, mental health

1 | I N T RO D U CT I O N difficulties, the data cause concern and justify further research (Ahola
et al., 2006).
Teachers' mental health has a widespread impact not just on teaching Although researchers have developed several instruments
effectiveness (Thakur, 2012) but even on students' personal and emo- (Kristensen, Borritz, Villadsen, & Christensen, 2005; Shirom &
tional development and academic achievement (Blazer, 2010; Ju, Lan, Melamed, 2006), the most widely used tool to asses burnout is the
Li, Feng, & You, 2015; Zhang & Sapp, 2008). Because teaching is a pro- Maslach Burnout Inventory (MBI; Maslach & Jackson, 1981). Maslach
fession with high mental risks (Kyriacou, 2001), several studies have and Jackson (1981) originally designed the MBI to assess burnout syn-
focused on teachers' well‐being and mental health (Parker, Martin, drome in a variety of human service occupations. Subsequently,
Colmar, & Liem, 2012; Van Droogenbroeck & Spruyt, 2015). Burnout Maslach and her colleagues have developed some alternative versions
is a psychological response to chronic, work‐related stress, character- for different sectors: The MBI–Human Services Survey (MBI–HSS) is
ized by emotional exhaustion, depersonalization, and feeling of low based on the original MBI and the human services sector uses it; users
personal accomplishment (Maslach, Schaufeli, & Leiter, 2001). Several in the education field apply the MBI–Educators Survey (MBI–ES;
studies have shown burnout's impact on health and productivity (Ahola Maslach, Jackson, & Schwab, 1996); and finally other professions use
et al., 2008; Dewa, Jacobs, Thanh, & Loong, 2014; Honkonen et al., the MBI–General Survey (MBI–GS). The MBI–ES is basically the same
2006; Mohren et al., 2003) including among health care workers as the MBI–HSS, with some minor changes in wording. For example,
(Consiglio, 2014; Pedersen, Sørensen, Bruun, Christensen, & Vedsted, the word “recipient” became “student.” The MBI–ES is intended to
2016; Potter, 2006; Ríos‐Risquez & García‐Izquierdo, 2016) and also measure three burnout dimensions: (a) the core aspect of burnout is
among teachers (Brunsting, Sreckovic, & Lane, 2014; Hoglund, Klingle, the emotional exhaustion, the chronic feeling of tiredness and fatigue,
& Hosan, 2015; Rudow, 1999). which make teachers lose their enthusiasm; (b) the second dimension is
The prevalence of burnout among teachers varies between 5 and depersonalization, which is a psychological withdrawal including the
30% depending on the conceptualization and measurement of burnout development of negative or indifferent attitudes toward students;
(Blazer, 2010; Gil‐Monte, Carlotto, & Gonçalves, 2011). Although it is and (c) the third dimension is the feeling of low personal accomplish-
difficult to determine the prevalence of burnout due to psychometric ment referring to losing the belief of one's effectiveness.

530 Copyright © 2017 John Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/smi Stress and Health. 2017;33:530–539.
SZIGETI ET AL. 531

Despite the wide use of the MBI, psychologists have not come to a as general severity and also to have a secondary loading on a specific
clear consensus regarding its factor structure. Several researchers have symptom dimension. Another advantage of adopting the bifactor
questioned the three‐factor structure Maslach and Jackson (1981) model is that researchers can study the contribution of the specific
originally specified. Most of the studies supported the original three‐ factors to predict an external variable independently from the general
factor model across different occupational groups (Bria, Spânu, Băban, factor. (Brunner, Nagy, & Wilhelm, 2012; Chen, West, & Sousa, 2006;
& Dumitraşcu, 2014; Gil Monte, 2005; Loera, Converso, & Viotti, Reise, Moore, & Haviland, 2010).
2014; Schutte, Toppinen, Kalimo, & Schaufeli, 2000), including As regards the MBI, in spite of its usefulness described above, the
teachers (Kokkinos, 2006; Silva, Hewage, & Fonseka, 2013); however, bifactor model was usually missing from the alternative models to test.
in order to support the original model, other users have requested Nevertheless, we ran a series of confirmatory factor analysis (CFA) in
various modifications, such as eliminating items, allowing items to load order to validate the Hungarian version of MBI–HSS in a healthcare pro-
on more than one factor, or allowing correlation between the factors fessional sample. In addition to the bifactor model, we also tested a
(Aluja, Blanch, & García, 2005; Gold, Roth, Wright, Michael, & Chin‐Yi, non‐hierarchical one‐factor, two‐factor, three‐factor, four‐factor, five‐
1992; Schaufeli, Daamen, & Van Mierlo, 1994). The results from a factor, and a hierarchical second‐order three‐factor model. The bifactor
meta‐analysis including 45 studies of confirmatory and exploratory model yielded the best fit to the data; however, only the general burn-
factor analyses, both descriptive and empirical analyses, showed sub- out factor and the specific personal accomplishment factor explained
stantial support for a correlated three‐factor model, but solutions with a considerable unique proportion of variance (Mészáros et al., 2014).
more or fewer factors also received modest support (Worley, Vassar, Due to the fact that there is no clear agreement regarding the
Wheeler, & Barnes, 2008). Some researchers have suggested fewer factorial structure of MBI–ES, it is still necessary to investigate the
dimensions in the burnout structure, such as combining emotional competing factor structures. The measurement model is essential
exhaustion and depersonalization into a single factor (Brookings, because it influences the interpretation of the three factor scores
Bolton, Brown, & McEvoy, 1985). Moreover, Kalliath, O'Driscoll, and general burnout score which can be especially important in
Gillespie, and Bluedorn (2000) excluded the personal accomplishment teacher population regarding they are less well studied compared to
factor yielding a two‐factor model consisting of just emotional exhaus- health workers. It is also not clear if burnout is a general construct
tion and depersonalization. Alternatively, some other researchers defined in the same way among teachers and among health workers
suggested an increased number of latent factors in order to explain a or if the burnout has some domain‐specific differences.
larger amount of variance. For example conceptualizing Maslach Burn- Symptoms of burnout and symptoms of depression are quite
out Inventory–General Survey as a five‐factor instrument, Densten similar; therefore, not surprisingly, numerous studies have reported a
(2001) further distinguished the psychological and somatic strain positive correlation between burnout and depressive symptoms (Ahola,
aspects of emotional exhaustion, and two aspects of the lack of per- Hakanen, Perhoniemi, & Mutanen, 2014; Ahola et al., 2005; Chiu,
sonal accomplishment, namely self and others. Emerging problems in Stewart, Woo, Yatham, & Lam, 2015; Iacovides, Fountoulakis, Kaprinis,
the studies on the factor structure of the MBI, like the large amount & Kaprinis, 2003; Takai et al., 2009). Considering the temporal and causal
of variance, remain unexplained (Aluja et al., 2005); the degree of the relationship of burnout and depressive symptoms, Shin, Noh, Jang, Park,
model's fitness is usually slightly lower than desired (Kokkinos, 2006); and Lee (2013) concluded that teachers' burnout can lead to subsequent
and the large correlations between the dimensions, especially between increases in depressive symptoms. Moreover, in some other studies,
emotional exhaustion and depersonalization (Worley et al., 2008), have researchers found that not only can burnout enhance depression, but
led to calls for an alternative measurement model. The presence of the also vice versa; depression can also enhance burnout, showing that these
strong correlation between the factors also implies a hierarchical problems are in a rather parallel/reciprocal relationship with each other
model containing a second order factor, as Worley et al. (2008) sug- (Bianchi, Schonfeld, & Laurent, 2015; Toker & Biron, 2012).
gested in their meta‐analysis. Because burnout and depression have some common characteris-
Previous researchers have proposed a hierarchical structure of tics, especially in the more severe forms of burnout, testing the diver-
psychological symptoms, and they have used two major types of gent validity of these two constructs is very important. Conceptually,
hierarchical models: second‐order factor models and bifactor models. burnout is job‐related whereas depression is context‐free and more per-
Although a vast majority of researchers in the field of personality and vasive, but this scope‐based distinction doesn't properly distinguish the
health research have been using the second order factor model for a late stages of burnout, which also pervasively affects non‐work areas as
long time, recently more have been using the bifactor model. In the well. A factor analysis has shown the two constructs to be statistically
second‐order factor models, the higher‐order factors explain the corre- differentiated from each other, where the items of burnout and depres-
lations or covariances of primary factors such as burnout factors. In the sion scales did not load on the same factor (Leiter & Durup, 1994).
bifactor model approach, one can estimate the general or common and Although the distinction still needs further clarification, all in all it
domain‐specific or unique components and their relative size in seems that depression and burnout can be regarded as related but
explaining variance. A bifactor measurement model specifies both a different constructs. Because of the overlap between the two
single common factor and at least two specific factors. Accordingly, constructs, there is a robust relationship between them. Therefore,
Reise (2012) proposed estimating the bifactor measurement structure depression is an appropriate construct to be used in testing the
as an effective approach to modeling construct‐relevant multidimen- construct validity of burnout measurement.
sionality. The bifactor measurement model allows for the indicators Another construct associated with burnout is overcommitment,
of psychological symptoms to load on an overall primary factor such which Siegrist et al. (2004) defined as an exaggerated involvement
532 SZIGETI ET AL.

and effort in work beyond what the organization formally requires, and pencil based. All participants gave written informed consent. The
which is a consequence of one's high feelings of need for esteem and Institutional Review Board of Institute of Psychology at Eötvös Loránd
approval. Overcommitted people are unable to withdraw from work University (Budapest, Hungary) approved the study.
obligations. Empirical studies have supported the theoretical assump- Two hundred eleven teachers (87.6% women) were willing to
tion that this intense continuous effort and the lack of ability to with- answer the questionnaire. The mean age was 42.8 years (SD = 10.5).
draw from work results in burnout, particularly with the fatigue/ The mean of the years the participants spent in the profession of
exhaustion component of burnout (Kristensen et al., 2005; Philp, Egan, teaching was 18.7 (SD = 11.0), and the mean years spent at the current
& Kane, 2012; Takaki, Nakao, Karita, Nishikitani, & Yano, 2006). How- workplace was 10.9 (SD = 9.8); 21.8% of the participants reported that
ever, the causal relationship of overcommitment and burnout is still their current workplace was also their first workplace in their lives. The
not clear. A current study with a longitudinal design pointed out that mean of the participants' working hours per week was 41.8 (SD = 10.9);
burnout predicts overcommitment as a maladaptive coping style and 20.7% of the participants stated that the majority of children they
an increased level of overcommitment in turn leads to more burnout teach was considered children with special needs, and for another
(Avanzi, Zaniboni, Balducci, & Fraccaroli, 2014). Accordingly, overcom- 28.8% of teachers, it was a frequent duty to take care of children with
mitment has a significant relationship with burnout, and therefore this special needs in their classes. The majority of participants (46.4%)
is also a relevant construct to be used in testing the construct validity worked in a primary school, 15.9% in secondary school, and 14.9% in
of burnout measurement. both.
The aim of this study was to test the construct validity of the
Hungarian version of MBI–ES. The first step in this process was to test
2.2 | Measurement tool
several alternative measurement models with CFA. The second step
was to investigate the associations of burnout factors with depressive 2.2.1 | Maslach Burnout Inventory–Education Survey
symptoms and overcommitment. We assessed burnout with the Hungarian version of the MBI–ES, which
To test the factor structure, we tested several previously proposed Maslach et al. (1996) specifically developed to measure burnout among
measurement models of the MBI–ES with CFA. We compared eight teachers. We obtained permission from the copyright owner (Mind
alternative models (see Figure 1) in which we also included a bifactor Garden, Inc.) to translate the MBI–ES into Hungarian language and to
model, which, despite its growing popularity, had not been tested pre- use it for the present study. We translated the instrument from English
viously with the Education Survey form. Finding the best fitting factor into Hungarian and validated by back‐translation to ensure that the two
structure of the measurement model is substantively important versions were equivalent. The questionnaire consists of 22 items. Emo-
because it influences the results' interpretation and application. For tional exhaustion is measured by nine items, including, for one example:
example, it can clarify whether one can use the global score of MBI– “I feel emotionally drained from my work.” Depersonalization consists
ES as an index of the underlying latent construct of general burnout. of five items, with a sample item being “I don't really care what happens
To support the construct validity of MBI–ES, we investigated the to some students.” Finally, personal accomplishment is formed by eight
associations of burnout factors with depressive symptoms and over- items, with a sample item of “I can easily create a relaxed atmosphere
commitment in one CFA with a covariates model. Regarding depres- with my students.” All items have a 7‐point response scale ranging from
sion, according to the literature, the strongest association emerged 0 (“never experienced such a feeling”) to 6 (“experience such feelings
with emotional exhaustion (Glass, McKnight, & Valdimarsdottir, 1993; every day”). Higher scores on the emotional exhaustion and deperson-
Tourigny, Baba, & Wang, 2010), which subscale is often considered to alization subscales indicate higher burnout, whereas higher scores on
be the core aspect of burnout. Because the general burnout factor cap- personal accomplishment indicate lower burnout. In the previous
tured by bifactor structure just reifies the core aspects of burnout, we research, the internal consistency of the factors ranged from 0.83 to
expected a more prominent relationship between depression and 0.91 for emotional exhaustion, from 0.50 to 0.83 for depersonalization,
general burnout than between depression and specific subscales. and from 0.69 to 0.82 for personal accomplishment. In the present
Overcommitment theoretically is related to one's being research, the internal consistencies were appropriate: 0.86 for
overwhelmed, investing too much energy in one's work. We expected emotional exhaustion, 0.64 for depersonalization, 0.76 for personal
that among the three components of MBI–ES, emotional exhaustion accomplishment, and 0.67 for general burnout.
would be most related to overcommitment, rather than depersonaliza-
tion or personal accomplishment, or even than general burnout. 2.2.2 | Center for Epidemiological Studies‐Depression
Scale
We used the Center for Epidemiological Studies‐Depression scale
(CES‐D; Radloff, 1977) to assess depressive symptoms, which instru-
2 | METHODS
ment Radloff (1977) developed to measure depressive symptomatol-
ogy in the general population. The measurement consists of 20 items
2.1 | Procedure and participants using a 4‐point frequency scale ranging from 0 (rarely or none of the
We based this cross‐sectional research on a convenient sample of time) to 3 (most or all of the time). The center for epidemiological stud-
teachers who were either living in Budapest or participating in a post- ies‐depression scale consists of a list of generic symptoms of depres-
graduate course in Budapest. Participation in the study was voluntary sion. We asked respondents to indicate how often they experienced
and anonymous. Administration of the questionnaires was paper and each described statement during the week that preceded the date of
SZIGETI ET AL. 533

FIGURE 1 The alternative measurement models of MBI–ES

the survey. A sample item is “I thought my life was a failure.” The inter- This is composed of six 4‐point Likert‐scaled items ranging from 0
nal consistency of the scale is high: 0.85 in the general population. In (strongly disagree) to 3 (strongly agree). Examples of these items
the present study, the internal consistency was good at 0.86. include “I get easily overwhelmed by time pressures at work” and
“I start thinking about work problems as soon as I get up in the
morning.” The total score varies between 0 and 18, where the higher
2.2.3 | Overcommitment (ERI‐Q) the score, the more likely a subject is to experience overcommitment
We assessed overcommitment using the Hungarian version of the at work. The Cronbach's alpha ranges from 0.64 to 0.82. In the
overcommitment scale (Salavecz, Neculai, Rózsa, & Kopp, 2006) from present study, the internal consistency was below the usual value
the Effort‐reward Imbalance Questionnaire (Siegrist et al., 2004). at 0.58.
534 SZIGETI ET AL.

2.2.4 | Working with children with special needs value above 0.10 indicates a poor fit. The closeness of the model
In order to measure the extent of working with children with special fit using RMSEA (CFit of RMSEA) is a statistical test (Browne &
needs, we used a Likert–type question such as “How typical is, that Cudeck, 1993) that evaluates the statistical deviation of RMSEA
you have to teach children with special needs as well (like dyslexia, from the value 0.05. Nonsignificant probability values (p > .05) indi-
ADHD, behavioral disorders, autism, etc.).” The possible answers cate an acceptable model fit (Brown, 2006). A satisfactory degree of
ranged from “It occurs very rarely, and even then is not very disturbing fit requires the comparative fit index (CFI) and the Tucker‐Lewis
(just a few classes where it occurs)” which has been coded as 1, to Index (TLI) to be higher than or close to 0.95 (Brown, 2006). The
“Mostly I teach these children (more than half of the class, or as work- next fit index is the standardized root mean square residual (SRMR).
ing in a special institution, every member of the class are with special An SRMR value below 0.08 is considered a good fit (Kline, 2011).
needs)” which has been coded as 4. The last fit index is the Bayesian information criteria, which do not
have a clear cut‐off. The general rule is that the lower number rep-
resents a closer fit.
3 | DATA ANALYSIS We also followed the double‐index strategy proposed by Hu and
Bentler (1999), who pointed out that a cut‐off value close to 0.08 for
We used structural equation modeling (SEM) with Mplus 6.0 to esti- SRMR and a cut‐off value close to 0.06 for RMSEA are necessary to
mate the degree of fit of eight alternative measurement models. The make a decision regarding a relatively good fit between the hypothe-
models were (a) a one‐factor model; (b) a two‐factor model sized model and the observed data. Hu and Bentler (1999) demon-
combining emotional exhaustion and depersonalization into one fac- strated that the combination of decision criteria including
tor (Brookings et al., 1985); (c) a two‐factor model with only emo- RMSEA ≤ 0.06 and SRMR ≤ 0.08 provides the opportunity to reject
tional exhaustion and depersonalization, eliminating personal misclassified models.
accomplishment (Kalliath et al., 2000); (d) a three‐factor model as In order to test the construct validity of the Hungarian version of
was originally proposed (Maslach & Jackson, 1981); (e) a second‐ MBI–ES, we tested a CFA with covariates model. Then, we ran a CFA
order factor model, with the three originally proposed factors and with the previously best fitting model with covariates. Covariates ana-
with a second‐order burnout factor; (f) a four‐factor model delineat- lyzed in the model were gender, age, depressive symptoms, and over-
ing job and client‐related aspects of depersonalization (Iwanicki & commitment. We entered the latter two variables as observed
Schwab, 1981); (g) a five‐factor model distinguishing two aspects variables.
of emotional exhaustion, namely somatic and psychological strain,
and two aspects of personal accomplishment, namely self and others
(Densten, 2001); and (h) a bifactor model, with the three factors and 4 | RESULTS
a general burnout factor.
We performed all analyses with maximum likelihood parameter We conducted a series of confirmatory factor analyses in order to find
estimates with standard errors and chi‐square test statistics that were the best fitting measurement model of the Hungarian version of the
robust to non‐normality of observation. We used the full information MBI–ES. The fit indices of the eight alternative measurement models
maximum likelihood estimator to deal with missing data (Muthén & are presented in Table 1. All models except Model 1 (one‐factor model)
Muthén, 2007). yielded SRMR value below 0.08. However, only the bifactor model
Because the chi‐square test is sensitive to sample size, where reached the low level of RMSEA that is not different significantly from
the probability of rejecting a hypothesized model increases with the predefined value of 0.05. Therefore, only model 8, the bifactor
the sample size, we used relative fit indices to analyze the models. model with a general burnout factor and three domain‐specific factors,
The first fit index was the root mean squared error of approximation yielded an acceptable level of model fit, suggesting that the bifactor
(RMSEA). An RMSEA below 0.05 indicates an excellent fit, and a model is superior to the alternative models.

TABLE 1 Confirmatory factor analysis of the eight measurement models of burnout

Alternative measurement models χ2 df RMSEA Cfit p CFI TLI SRMR BIC

Model 1. One‐factor model 547** 209 0.088 <0.001 0.70 0.66 0.089 15719
Model 2.Two‐factor model (Brookings et al., 1985) 442** 208 0.073 <0.001 0.79 0.77 0.078 15592
Model 3. Two factor model (Kalliath et al., 2000) 200** 76 0.088 <0.001 0.82 0.79 0.070 10089
Model 4. Three‐factor model (Maslach & Jackson, 1981) 432** 206 0.072 <0.001 0.80 0.77 0.076 15586
Model 5. Hierarchical Second‐order model with three factor 432** 206 0.072 <0.001 0.80 0.77 0.076 15586
Model 6. Four‐factor model (Iwanicki & Schwab, 1981) 423** 203 0.072 <0.001 0.80 0.77 0.074 15589
Model 7. Five‐factor model (Densten, 2001; Iwanicki & Schwab, 1981) 284** 142 0.069 <0.001 0.82 0.79 0.071 13368
Model 8. Bifactor model (Mészáros et al., 2014) 302** 189 0.053 0.30 0.90 0.86 0.058 15515

Notes. χ = chi‐square test statistic; df = degrees of freedom;

2

**: p < .001; CFI = comparative fit index; TLI = Tucker–Lewis index; RMSEA = root mean square error of approximation; Cfit = closeness of fit. SRMR = stan-
dardized root mean square residual.
SZIGETI ET AL. 535

The standardized factor loadings for the bifactor model are We also calculated the explained common variances (Sijtsma,
shown in Table 2. All items except item 4 loaded significantly on 2009) and found that general burnout explained 58% of the common
the general burnout factor. The items of personal accomplishment variance, while the proportions of explained common variance by the
had negative factor loadings on this general burnout factor, being a specific factors were 13% for emotional exhaustion, 11% for deper-
positive measurement versus the other two specific factors. All eight sonalization, and 17% for personal accomplishment. This result also
items of personal accomplishment had significant factor loadings supports the bifactor structure, showing that the general factor
whereas not all items loaded significantly on their specific factor in explains a bigger amount of variance of the items than do the specific
the case of emotional exhaustion and depersonalization. Regarding factors all together.
emotional exhaustion, most of the items were loaded significantly: We evaluated the internal consistency by calculating omega and
items about “feeling used up,” “feeling fatigued,” and “working too omega hierarchical coefficients, besides Cronbach's alpha, which are
hard” with high factor loadings and items about “feeling emotionally recommended in case of multidimensionality and hierarchically struc-
drained,” “working with people is a strain,” and “feeling frustrated” tured constructs (Zinbarg, Revelle, Yovel, & Li, 2005) because
with lower factor loadings, the remaining three items about “feeling Cronbach alpha can be a misleading index in these cases (Cortina,
burned out,” “working with people is stressful” and “feeling at the 1993). Omega coefficients estimate how a certain scale score assesses
end of my rope” did not load significantly on this specific factor, the combination of general and specific constructs, and a certain target
but did so on the global factor. From the five items of depersonaliza- construct (Brunner et al., 2012). Omega coefficients estimate the
tion, two items loaded significantly: “seeing students as impersonal reliability of a latent factor (such as emotional exhaustion score) com-
objects,” “becoming more callous,” and the remaining three items: bining the general and the specific factor variance. Therefore, the
“emotionally hardening,” “not taking care of students' and “students omega coefficients of emotional exhaustion show the proportion of
blame me” did not load significantly on this factor, but did on the variance attributable to a mixture of the general burnout factor and
general factor. the specific emotional exhaustion factor. Contrarily, the coefficient

TABLE 2 Standardized factor loadings in the bifactor model of MBI–ES

Short form of itemsc Emotional exhaustion Depersonalization Personal accomplishment General burnout

1. feel emotionally drained 0.281 0.646

2. feel used up 0.871 0.490
3. feel fatigued 0.477 0.488
6. working with people is a strain 0.148 0.526
8. feel burned out 0.042 0.722
13. feel frustrated 0.113 0.780
14. working too hard 0.352 0.394
16. working with people is stressful 0.077 0.617
20. feel at the end of one's rope −0.054 0.685
5. students as impersonal objects 0.357 0.321
10. become more callous 0.947 0.476
11. emotionally hardening 0.009 0.571
15. not taking care of students 0.140 0.312
22. students blame me 0.104 0.560
4. understanding students 0.382 −0.026
7. dealing effectively with student's problems 0.405 −0.379
9. positive influence on others 0.538 −0.231
12. feel energetic 0.337 −0.507
17. creating relaxed atmosphere 0.420 −0.279
18. feel exhilarated with students 0.526 −0.350
19. accomplishing worthwhile things 0.545 −0.390
21. deal with emotional problems calmly 0.342 −0.394
Cronbach α 0.86 0.64 0.76 0.67
Omegaa 0.88 0.72 0.77 0.90
Omega hierarchicalb 0.15 0.23 0.50 0.75

Notes. Boldfaced factor loadings are significant at, at least p < .05.
a
Omega refers to the proportion of explained variance in the scale score attributed to the global and specific factors.
b
: Omega hierarchical refers to the proportion of explained variance of the scale score attributed to the specific factor.
Short form of items
c
because the MBI–ES is a copyrighted measurement tool, we used the main phrases instead of the whole sentences.
536 SZIGETI ET AL.

omega hierarchical estimates the reliability of a latent factor with all 5 | DISCUSSI ON AND I M P LI CATIONS
other latent construct variances removed. For example, in this case,
the coefficient omega hierarchical of emotional exhaustion shows the The present study supported the bifactor measurement model of MBI
proportion of variance of this scale that is attributable to this specific in a sample of teachers similarly to our previous study in a sample of
factor only, without the general burnout factor. Thus, the omega health care workers (Mészáros et al., 2014). Furthermore, we found
hierarchical provides information on whether scores for a specific that emotional exhaustion and depersonalization scores mainly reflect
factor can be interpreted per se or if only the general factor score general burnout rather than their specific meanings, and only a small
can be used (Brunner et al., 2012; Reise et al., 2010). amount of variance is explained by the specific factors. However, the
We calculated the coefficients for the best fitting bifactor model score of personal accomplishment reflects the specific meaning in a
for both the general and the specific factors, and these are presented higher degree, which is similar to our previous study (Mészáros et al.,
in Table 2. Omega coefficients were appropriate; thus, the specific 2014).
scale scores adequately represent the mixture of general burnout and We also estimated the covariates of burnout factors, namely
the respective specific factors. extent of teaching children with special needs, depressive symptoms
There is no clearly defined cut‐off for omega hierarchical coeffi- and overcommitment. The extent of teaching children with special
cients. However, Reise, Bonifay, and Haviland (2013) tentatively pro- needs was significantly associated only with the general factor. This
posed rather strict advice to evaluate omega hierarchical values for implies that this external impeding factor has influence on the overall
specific factors, specifically that the minimum value would be greater burnout phenomenon, and does not have effect through specific fac-
than 0.50 and the preferred value would be closer to 0.75. Because tors such as having little success therefore having less personal accom-
values of the omega hierarchical also depend on such factors as the plishment, or through having greater emotional burden. Depressive
number of specific factors and the test length (Reise et al., 2013), we symptoms were significantly associated only with the general burnout
proposed a more lenient approach in this research. We regarded coef- factor. This finding is in accordance with the previously reported
ficients around 0.20 as salient, in which case the specific factor would robust connection between burnout and depression (Ahola et al.,
explain at least 20% of the variance in specific factor scores. Using our 2014; Glass et al., 1993; Tourigny et al., 2010). Furthermore, this study
criterion, the contribution of a specific factor to the explained variance also highlighted the fact that the depressive symptoms are related to
would not be negligible. Regarding the omega hierarchical, the emo- the general aspect of burnout, and this association is not limited to
tional exhaustion and the depersonalization had quite a low value, any specific dimension. Overcommitment was significantly associated
implying that only a small proportion of the observed variance is attrib- with the general burnout and emotional exhaustion dimensions.
utable to the respective component, and the general burnout factor Although the link between overcommitment and burnout was
explains a much larger proportion of the variance in these scores. Per- previously repeated (Avanzi et al., 2014; Philp et al., 2012), our study
sonal accomplishment has an omega hierarchical coefficient with a could present evidence that controlling for the link between overcom-
moderate value of around 0.50, denoting that its score is attributable mitment and general burnout factor, overcommitment is also related to
partly to the specific underlying dimension and partly to the general the emotional exhaustion dimension. This finding is in line with the
burnout factor. previous studies in which burnout was measured by the work‐related
In order to test the concurrent validity of MBI–ES, we ran the factor of the Copenhagen Burnout Inventory (Avanzi et al., 2014;
bifactor model with covariates including depression, and overcommit- Philp et al., 2012) or by fatigue (Takaki et al., 2006), which items show
ment controlling for gender and age. We also introduced in the model similarity in content with the items of emotional exhaustion.
the extent of teaching children with special need as another covariate. Our study has some important implications. The results suggest
The standardized coefficients are presented in Table 3. Depressive that in further studies, MBI–ES should be used as a bifactor measure-
symptoms and the extent of teaching children with special needs had ment with its advantages (Reise, 2012), using the latent variable
a significant and strong relationship with the general burnout factor, approach, both the general and specific components can be used to
but had no significant relationship with the specific subscales. Over- gain a better understanding of burnout. Our study also suggests that
commitment was in a significant relationship with the emotional in the conceptualization of burnout, unidimensionality could be a more
exhaustion factor and also with the general burnout factor. appropriate approach. The scores of emotional exhaustion and
probably depersonalization should be used cautiously because they
TABLE 3 Standardized coefficients of CFA with covariates model may reflect more the general burnout dimension and only a small
portion of variance could be attributed to their specific meanings.
Special Depressive
needsa symptoms Overcommitment R2 Another implication of this study is that teaching children with special

Emotional exhaustion −0.04 0.19 0.31 20% needs and depressive symptoms are closely related to the general

Depersonalization 0.01 −0.02 −0.11 2% burnout factor and not to the specific factors. However, due to the

Personal −0.06 −0.03 0.15 9% cross‐sectional nature of this study, it is not clear if burnout increases
accomplishment the depressive symptoms or if teachers having the tendency to
General burnout 0.15 0.58 0.23 56% experience depressive symptoms are more likely to suffer more

Notes. Boldfaced regression coefficients are significant at, at least p < .05; burnout symptoms. Clarification of the proper use of this measure-
special needs ment is a fundamental contribution to helping the prevention, the early
a
means the extent of teaching children with special needs. detection, and the treatment of burnout.
SZIGETI ET AL. 537

We should also mention this study's limitations. First, the cross‐ Romanian healthcare professionals. Burnout Research, 1(3), 103–111.
sectional design does not allow us to make assumptions about the doi:10.1016/j.burn.2014.09.001

causality. Second, being a self‐report study with paper‐ and pencil‐ Brookings, J. B., Bolton, B., Brown, C. E., & McEvoy, A. (1985). Self‐
reported job burnout among female human service professionals.
based questionnaires, it is not free from several biases. And third,
Journal of Organizational Behavior, 6(2), 143–150. doi:10.1002/
having a convenient sample, we could not have a representative data job.4030060205
of Hungarian teachers. Even though the sample consisted of teachers Brown, T. A. (2006). Confirmatory factor analysis for applied research. New
from different school types (elementary and high school), the York: Guilford.
generalizability of the results is limited. Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing
model fit. In K. A. L. Bollen & J. S. Long (Ed.), Testing structural equa-
tion models. (pp. 136–162). Newbury Park: Sage.
6 | C O N CL U S I O N S Brunner, M., Nagy, G., & Wilhelm, O. (2012). A Tutorial on hierarchically
structured constructs. Journal of Personality, 80(4), 796–846.
Our study revealed the bifactor structure of MBI–ES and thus Brunsting, N. C., Sreckovic, M. A., & Lane, K. L. (2014). Special education
proposes a more sophisticated analysis of the results of the scale in teacher burnout: A synthesis of research from 1979 to 2013. Education
& Treatment of Children, 37(4), 681–711.
further research and also encourages the further unidimensional
Chen, F. F., West, S. G., & Sousa, K. H. (2006). A comparison of bifactor and
conceptualization of burnout. Teaching children with special needs,
second‐order models of quality of life. Multivariate Behavioral Research,
depressive symptoms and overcommitment are connected with the 41(2), 189–225. doi:10.1207/s15327906mbr4102_5
general aspects of burnout. Additionally, unlike depressive symptoms, Chiu, L. Y. L., Stewart, K., Woo, C., Yatham, L. N., & Lam, R. W. (2015). The
overcommitment has a stronger relationship with one specific compo- relationship between burnout and depressive symptoms in patients
nent of burnout, namely emotional exhaustion. It is still not clear if with depressive disorders. Journal of Affective Disorders, 172, 361–
366. doi:10.1016/j.jad.2014.10.029
depressive symptoms increase burnout or if burnout can increase
Consiglio, C. (2014). Interpersonal strain at work: A new burnout facet rel-
depressive symptoms. They may also develop in parallel. Further
evant for the health of hospital staff. Burnout Research, 1(2), 69–75.
research is necessary to identify the causal pathways in order to plan doi:10.1016/j.burn.2014.07.002
appropriate prevention programs for teachers.
Cortina, J. M. (1993). What is coefficient alpha? An examination of theory
and applications. Journal of Applied Psychology, 78(1), 98–104.
CONF LICT OF INTE R ES T doi:10.1037/0021-9010.78.1.98
Densten, I. L. (2001). Re‐thinking burnout. Journal of Organizational
The authors have declared that they have no conflict of interest.
Behavior, 22(8), 833–847. doi:10.1002/job.115
Dewa, C. S., Jacobs, P., Thanh, N. X., & Loong, D. (2014). An estimate of the
RE FE R ENC E S
cost of burnout on early retirement and reduction in clinical hours of
Ahola, K., Hakanen, J., Perhoniemi, R., & Mutanen, P. (2014). Relationship practicing physicians in Canada. BMC Health Services Research, 14(1),
between burnout and depressive symptoms: A study using the 1–9. doi:10.1186/1472-6963-14-254
person‐centred approach. Burnout Research, 1(1), 29–37. doi:10.1016/
j.burn.2014.03.003 Gil‐Monte, P. R., Carlotto, M. S., & Gonçalves, C. S. (2011). Prevalence of
burnout in a sample of Brazilian teachers. The European Journal of
Ahola, K., Honkonen, T., Isometsä, E., Kalimo, R., Nykyri, E., Aromaa, A., & Psychiatry, 25, 205–212.
Lönnqvist, J. (2005). The relationship between job‐related burnout
and depressive disorders—Results from the Finnish Health 2000 Study. Gil Monte, P. P. (2005). Factorial validity of the Maslach Burnout Inven-
Journal of Affective Disorders, 88(1), 55–62. doi:10.1016/j. tory (MBI‐HSS) among Spanish professionals. Revista de Saúde
jad.2005.06.004 Pública, 39, 1–8.

Ahola, K., Honkonen, T., Isometsa, E., Kalimo, R., Nykyri, E., Koskinen, S., … Glass, D. C., McKnight, J. D., & Valdimarsdottir, H. (1993). Depression,
Lönnqvist, J. (2006). Burnout in the general population. Results from burnout, and perceptions of control in hospital nurses. Journal of Con-
the Finnish Health 2000 Study. Social Psychiatry and Psychiatric sulting and Clinical Psychology, 61(1), 147–155. doi:10.1037/0022-
Epidemiology, 41(1), 11–17. doi:10.1007/s00127-005-0011-5 006X.61.1.147

Ahola, K., Kivimaki, M., Honkonen, T., Virtanen, M., Koskinen, S., Vahtera, Gold, Y., Roth, R. A., Wright, C. R., Michael, W. B., & Chin‐Yi, C.
J., & Lonnqvist, J. (2008). Occupational burnout and medically certified (1992). The factorial validity of a teacher burnout measure
sickness absence: A population‐based study of Finnish employees. (educators survey) administered to a sample of beginning teachers
Journal of Psychosomatic Research, 64(2), 185–193. doi:10.1016/j. in elementary and secondary schools in California. Educational and
jpsychores.2007.06.022 Psychological Measurement, 52(3), 761–768. doi:10.1177/
0013164492052003027
Aluja, A., Blanch, A., & García, L. F. (2005). Dimensionality of the Maslach
Burnout Inventory in school teachers. European Journal of Psychological Hoglund, W. L. G., Klingle, K. E., & Hosan, N. E. (2015). Classroom risks and
Assessment, 21(1), 67–76. doi:10.1027/1015-5759.21.1.67 resources: Teacher burnout, classroom quality and children's adjust-
ment in high needs elementary schools. Journal of School Psychology,
Avanzi, L., Zaniboni, S., Balducci, C., & Fraccaroli, F. (2014). The relation 53(5), 337–357. doi:10.1016/j.jsp.2015.06.002
between overcommitment and burnout: Does it depend on employee
job satisfaction? Anxiety, Stress, & Coping, 27(4), 455–465. Honkonen, T., Ahola, K., Pertovaara, M., Isometsa, E., Kalimo, R., Nykyri, E.,
doi:10.1080/10615806.2013.866230 … Lönnqvist, J. (2006). The association between burnout and physical
illness in the general population‐‐results from the Finnish Health
Bianchi, R., Schonfeld, I. S., & Laurent, E. (2015). Burnout–depression over- 2000 Study. Journal of Psychosomatic Research, 61(1), 59–66.
lap: A review. Clinical Psychology Review, 36, 28–41. doi:10.1016/j. doi:10.1016/j.jpsychores.2005.10.002
cpr.2015.01.004
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance
Blazer, C. (2010). Teacher burnout. Information Capsule, 1003.
structure analysis: Conventional criteria versus new alternatives. Struc-
Bria, M., Spânu, F., Băban, A., & Dumitraşcu, D. L. (2014). Maslach Burnout tural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55.
Inventory – General Survey: Factorial validity and invariance among doi:10.1080/10705519909540118
538 SZIGETI ET AL.

Iacovides, A., Fountoulakis, K. N., Kaprinis, S., & Kaprinis, G. (2003). The Australian Journal of Psychology, 64(2), 68–74. doi:10.1111/j.1742-
relationship between job stress, burnout and clinical depression. 9536.2011.00028.x
Journal of Affective Disorders, 75(3), 209–221. doi:10.1016/S0165- Potter, C. (2006). To what extent do nurses and physicians working within
0327(02)00101-5 the emergency department experience burnout: A review of the litera-
Iwanicki, E. F., & Schwab, R. L. (1981). A cross validation study of the ture. Australasian Emergency Nursing Journal, 9(2), 57–64. doi:10.1016/j.
Maslach Burnout Inventory. Educational and Psychological Measurement, aenj.2006.03.006
41(4), 1167–1174. doi:10.1177/001316448104100425 Radloff, L. S. (1977). The CES‐D Scale: A self‐report depression scale for
Ju, C., Lan, J., Li, Y., Feng, W., & You, X. (2015). The mediating role of research in the general population. Applied Psychological Measurement,
workplace social support on the relationship between trait emotional 1(3), 385–401. doi:10.1177/014662167700100306
intelligence and teacher burnout. Teaching and Teacher Education, 51, Reise, S. P. (2012). The rediscovery of bifactor measurement models. Mul-
58–67. tivariate Behavioral Research, 47(5), 667–696. doi:10.1080/
Kalliath, T. J., O'Driscoll, M. P., Gillespie, D. F., & Bluedorn, A. C. (2000). A 00273171.2012.715555
test of the Maslach Burnout Inventory in three samples of healthcare Reise, S. P., Bonifay, W. E., & Haviland, M. G. (2013). Scoring and model-
professionals. Work & Stress, 14(1), 35–50. doi:10.1080/ ing psychological measures in the presence of multidimensionality.
026783700417212 Journal of Personality Assessment, 95(2), 129–140. doi:10.1080/
Kline, R. B. (2011). Principles and practice of structural equation modeling 00223891.2012.725437
(3rd ed.). New York: The Guilford Press. Reise, S. P., Moore, T. M., & Haviland, M. G. (2010). Bifactor models and
rotations: Exploring the extent to which multidimensional data yield
Kokkinos, C. M. (2006). Factor structure and psychometric properties of
univocal scale scores. Journal of Personality Assessment, 92(6), 544–
the Maslach Burnout Inventory‐Educators Survey among elementary
559. doi:10.1080/00223891.2010.496477
and secondary school teachers in Cyprus. Stress and Health, 22(1),
25–33. doi:10.1002/smi.1079 Ríos‐Risquez, M. I., & García‐Izquierdo, M. (2016). Patient satisfaction,
stress and burnout in nursing personnel in emergency departments: A
Kristensen, T. S., Borritz, M., Villadsen, E., & Christensen, K. B. (2005). The
cross‐sectional study. International Journal of Nursing Studies, 59, 60–
Copenhagen Burnout Inventory: A new tool for the assessment of
67. doi:10.1016/j.ijnurstu.2016.02.008
burnout. Work & Stress, 19(3), 192–207. doi:10.1080/
02678370500297720 Rudow, B. (1999). Stress and burnout in the teaching profession: European
studies, issues, and research perspectives. In R. V. A. M. Huberman
Kyriacou, C. (2001). Teacher stress: Directions for future research. (Ed.), Understanding and preventing teacher burnout: A sourcebook of
Educational Review, 53(1), 27–35. doi:10.1080/00131910120033628 international research and practice. (pp. 38–58). New York, NY, US:
Leiter, M. P., & Durup, J. (1994). The discriminant validity of burnout and Cambridge University Press.
depression: A confirmatory factor analytic study. Anxiety, Stress, & Salavecz, G., Neculai, K., Rózsa, S., & Kopp, M. (2006). Az erőfeszítés‐
Coping, 7(4), 357–373. doi:10.1080/10615809408249357 jutalom egyensúlytalanság kérdőív magyar változatának
Loera, B., Converso, D., & Viotti, S. (2014). Evaluating the psychometric megbízhatósága és érvényesség. Mentálhigiéné és Pszichoszomatika,
properties of the Maslach Burnout Inventory‐Human Services Survey 7(3), 231–246.
(MBI‐HSS) among Italian Nurses: How Many Factors Must a Schaufeli, W. B., Daamen, J., & Van Mierlo, H. (1994). Burnout among
Researcher Consider? PloS One, 9(12) e114987. doi:10.1371/journal. Dutch teachers: An Mbi‐validity study. Educational and Psychological
pone.0114987 Measurement, 54(3), 803–812. doi:10.1177/0013164494054003027
Maslach, C., & Jackson, S. E. (1981). The measurement of experienced Schutte, N., Toppinen, S., Kalimo, R., & Schaufeli, W. (2000). The
burnout. Journal of Occupational Behaviour, 2, 99–113. doi:10.1002/ factorial validity of the Maslach Burnout Inventory‐General Survey
job.4030020205 (MBI‐GS) across occupational groups and nations. Journal of Occupa-
Maslach, C., Jackson, S. E., & Schwab, R. L. (1996). Maslach Burnout Inven- tional and Organizational Psychology, 73(1), 53–66. doi:10.1348/
tory‐Educators Survey (MBI‐ES). In C. Maslach, S. E. Jackson, & M. P. 096317900166877
Leiter (Eds.), MBI Manual (3rd ed.)). Palo Alto, CA: Consulting Psycholo- Shin, H., Noh, H., Jang, Y., Park, Y. M., & Lee, S. M. (2013). A longitudinal
gists Press. examination of the relationship between teacher burnout and depres-
sion. Journal of Employment Counseling, 50(3), 124–137. doi:10.1002/
Maslach, C., Schaufeli, W. B., & Leiter, M. P. (2001). Job burnout. Annual
j.2161-1920.2013.00031.x
Review of Psychology, 52, 397–422.
Shirom, A., & Melamed, S. (2006). A comparison of the construct validity of
Mészáros, V., Adám, S., Szabó, M., Szigeti, R., & Urbán, R. (2014). The
two burnout measures in two groups of professionals. International
Bifactor Model of the Maslach Burnout Inventory‐Human Services Sur-
Journal of Stress Management, 13(2), 176–200.
vey (MBI‐HSS)‐An Alternative Measurement Model of Burnout Stress
and Health, 30(1), 82–88. Siegrist, J., Starke, D., Chandola, T., Godin, I., Marmot, M., Niedhammer, I., &
Peter, R. (2004). The measurement of effort–reward imbalance at work:
Mohren, D. C. L., Swaen, G. M. H., Kant, I., van Amelsvoort, L. G. P. M., European comparisons. Social Science & Medicine, 58(8), 1483–1499.
Borm, P. J. A., & Galama, J. M. D. (2003). Common infections and the
doi:10.1016/S0277-9536(03)00351-4
role of burnout in a Dutch working population. Journal of Psychosomatic
Research, 55(3), 201–208. doi:10.1016/S0022-3999(02)00517-2 Sijtsma, K. (2009). On the use, the misuse, and the very limited usefulness
of cronbach's alpha. Psychometrika, 74(1), 107–120. doi:10.1007/
Muthén, L. K., & Muthén, B. O. (2007). Mplus user's guide (Fifth ed.). Los s11336-008-9101-0
Angeles: Muthen & Muthen.
Silva, P. D., Hewage, C., & Fonseka, P. (2013). Sinhala translation of
Parker, P. D., Martin, A. J., Colmar, S., & Liem, G. A. (2012). Teachers' work- Maslach Burnout Inventory ‐ Educators Survey among female primary
place well‐being: Exploring a process model of goal orientation, coping school teachers in Sri Lanka. Galle Medical Journal, 18(1), 8–11.
behavior, engagement, and burnout. Teaching and Teacher Education, doi:10.4038/gmj.v18i1.5512
28(4), 503–513. doi:10.1016/j.tate.2012.01.001
Takai, M., Takahashi, M., Iwamitsu, Y., Ando, N., Okazaki, S., Nakajima, K., …
Pedersen, A. F., Sørensen, J. K., Bruun, N. H., Christensen, B., & Vedsted, P. Miyaoka, H. (2009). The experience of burnout among home caregivers
(2016). Risky alcohol use in Danish physicians: Associated with of patients with dementia: Relations to depression and quality of life.
alexithymia and burnout? Drug and Alcohol Dependence, 160, 119– Archives of Gerontology and Geriatrics, 49(1), e1–e5. doi:10.1016/j.
126. doi:10.1016/j.drugalcdep.2015.12.038 archger.2008.07.002
Philp, M., Egan, S., & Kane, R. (2012). Perfectionism, over commitment to Takaki, J., Nakao, M., Karita, K., Nishikitani, M., & Yano, E. (2006). Relation-
work, and burnout in employees seeking workplace counselling. ships between effort‐reward imbalance, over‐commitment, and fatigue
SZIGETI ET AL. 539

in Japanese information‐technology workers. Journal of Occupational studies. Educational and Psychological Measurement, 68(5), 797–823.
Health, 48(1), 62–64. doi:10.1539/joh.48.62 doi:10.1177/0013164408315268
Thakur, S. S. (2012). To study the relationship between burnout and Zhang, Q., & Sapp, D. A. (2008). A burning issue in teaching: The impact of
effectiveness of primary school teachers. International Indexed & teacher burnout and nonverbal immediacy on student motivation and
Referred Journal, 3(30), 15–16. affective learning. Journal of Communication Studies, 1(2), 152–168.
Toker, S., & Biron, M. (2012). Job burnout and depression: Unraveling their Zinbarg, R. E., Revelle, W., Yovel, I., & Li, W. (2005). Cronbach's α, Revelle's
temporal relationship and considering the role of physical activity. Jour- β, and Mcdonald's ω H : Their relations with each other and two alter-
nal of Applied Psychology, 97(3), 699. native conceptualizations of reliability. Psychometrika, 70(1), 123–133.
Tourigny, L., Baba, V. V., & Wang, X. (2010). Burnout and depression among doi:10.1007/s11336-003-0974-7
nurses in Japan and China: The moderating effects of job satisfaction
and absence. The International Journal of Human Resource Management,
21(15), 2741–2761. doi:10.1080/09585192.2010.528656 How to cite this article: Szigeti R, Balázs N, Bikfalvi R, Urbán
Van Droogenbroeck, F., & Spruyt, B. (2015). Do teachers have worse men- R. Burnout and depressive symptoms in teachers: Factor struc-
tal health? Review of the existing comparative research and results ture and construct validity of the Maslach Burnout inventory‐
from the Belgian Health Interview Survey. Teaching and Teacher Educa-
educators survey among elementary and secondary school
tion, 51, 88–100. doi:10.1016/j.tate.2015.06.006
teachers in Hungary. Stress and Health. 2017;33:530–539.
Worley, J. A., Vassar, M., Wheeler, D. L., & Barnes, L. L. B. (2008). Factor
structure of scores from the Maslach Burnout Inventory: A review https://doi.org/10.1002/smi.2737
and meta‐analysis of 45 exploratory and confirmatory factor‐analytic
Copyright of Stress & Health: Journal of the International Society for the Investigation of
Stress is the property of John Wiley & Sons, Inc. and its content may not be copied or emailed
to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.