ORIGINAL RESEARCH ARTICLE

published: 11 December 2014

doi: 10.3389/fpsyg.2014.01442

Teachers’ emotional experiences and exhaustion as

predictors of emotional labor in the classroom: an
experience sampling study
Melanie M. Keller 1,2 *, Mei-Lin Chang 3 , Eva S. Becker 1,2 , Thomas Goetz1,2 and Anne C. Frenzel 4
1
Department of Empirical Educational Research, University of Konstanz, Konstanz, Germany
2
Department of Empirical Educational Research, Thurgau University of Teacher Education, Kreuzlingen, Switzerland
3
Secondary and Middle Grades Education, Kennesaw State University, Atlanta, GA, USA
4
Department of Psychology, University of Munich, Munich, Germany

Edited by: Emotional exhaustion (EE) is the core component in the study of teacher burnout, with
Barbara McCombs, University of significant impact on teachers’ professional lives. Yet, its relation to teachers’ emotional
Denver, USA
experiences and emotional labor (EL) during instruction remains unclear.Thirty-nine German
Reviewed by:
Katrina Liu, The University of
secondary teachers were surveyed about their EE (trait), and via the experience sampling
Wisconsin-Whitewater, USA method on their momentary (state; N = 794) emotional experiences (enjoyment, anxiety,
Sung-il Kim, Korea University, South anger) and momentary EL (suppression, faking).Teachers reported that in 99 and 39% of all
Korea lessons, they experienced enjoyment and anger, respectively, whereas they experienced
*Correspondence: anxiety less frequently. Teachers reported suppressing or faking their emotions during
Melanie M. Keller, Department of
Empirical Educational Research,
roughly a third of all lessons. Furthermore, EE was reflected in teachers’ decreased
University of Konstanz, experiences of enjoyment and increased experiences of anger. On an intra-individual level,
Universitaetsstrasse 10, Fach 45, all three emotions predict EL, whereas on an inter-individual level, only anger evokes EL.
Konstanz, Germany Explained variances in EL (within: 39%, between: 67%) stress the relevance of emotions
e-mail: melanie.keller@
uni-konstanz.de
in teaching and within the context of teacher burnout. Beyond implying the importance of
reducing anger, our findings suggest the potential of enjoyment lessening EL and thereby
reducing teacher burnout.
Keywords: teacher emotions, teacher emotional labor, teacher emotional exhaustion, experience sampling method,
intra-individual vs. inter-individual analyses

INTRODUCTION factors in explaining EE (Morris and Feldman, 1996; Abraham,

It has been recognized that being a teacher is a demanding and 1999; Chang, 2009, 2013). However, only a few studies have estab-
sometimes even exhausting job. High dropout rates and the early lished empirical relationships between these factors in studying
retirement of teachers (see Macdonald, 1999) have caused some teacher emotions (Carson, 2006; Chang, 2009), many of which
societal alarm in recent years, prompting studies focusing on have relied on cross-sectional and one-time survey data (e.g.,
teacher burnout as a potential cause for teacher attrition (Chang, Hakanen et al., 2006; Skaalvik and Skaalvik, 2007). In Emotion
2009; see also Ashforth and Lee, 1990). Compared to other pro- in Education, Schutz and Pekrun (2007) argued for the need to
fessions, teaching in fact poses a relatively high risk of burnout study emotions in real-life contexts and to use multi-method
(de Heus and Diekstra, 1999; Brouwers and Tomic, 2000; see also, approaches, so that the complexities of emotional processes could
Maslach et al., 2001). Burnout, defined as “a psychological syn- be fully understood. In the present study, we examine the links
drome in response to chronic interpersonal stressors on the job” among teacher emotions, EL, and EE with momentary data,
(Maslach et al., 2001, p. 399), is conceptualized in scientific studies utilizing the experience sampling method (ESM).
via three dimensions: emotional exhaustion (EE), depersonaliza-
tion, and reduced personal accomplishment. EE, considered the RELEVANCE OF TEACHERS’ EMOTIONS TO EMOTIONAL EXHAUSTION
core facet of burnout (see for example Maslach et al., 2001; Chang, Being a teacher, and teaching in particular, is described as an
2009), refers to having depleted one’s emotional resources and emotional practice (Hargreaves, 1998), and emotions are charac-
therefore feeling emotionally overextended (Evers et al., 2004). terized as being “an integral part of teachers’ lives” (Sutton and
Beyond being related to teachers’motivation, for example job satis- Wheatley, 2003, p. 332). However, scientific studies of teach-
faction (Wolpin et al., 1991) or enthusiasm (Kunter et al., 2011), EE ers’ emotions have only surfaced within the last 15 years. Since
has also been shown to impact teaching quality (Klusmann et al., then, it has been established that teachers experience a variety of
2008). Thus, teacher burnout is a relevant factor in the study of discrete emotions in the course of their professional lives, partic-
teachers’ professional lives; yet our understanding of the emotional ularly while delivering instruction (e.g., Nias, 1996; Keller et al.,
processes in the classroom is still limited (Chang, 2009). 2014). Emotions are thought to be predictors of teacher behav-
Besides workload and lack of resources, emotional labor (EL) ior in class, in terms of effective instructional practices, as well as
and negative emotions have also been found to be contributing student behavior and outcomes (see theoretical model in Frenzel,

www.frontiersin.org December 2014 | Volume 5 | Article 1442 | 1

Keller et al. Teachers’ emotional labor during class

2014). Emotions are also relevant within the context of teachers’ expressive suppression and their EE. More specifically, when teach-
health and psychological well-being (for a general discussion, see ers reported engaging in expressive suppression, they also reported
Fredrickson, 1998). experiencing increased levels of EE. These results are consistent
Outside the teaching profession, there is evidence that with those of Brotheridge and Grandey (2002), who found a sig-
burnout – in particular EE – is strongly associated with increased nificant relationship between surface acting (e.g., hiding anger and
negative affectivity; or conversely, decreased positive affectivity fear) and EE.
(e.g., Brotheridge and Grandey, 2002). However, in teacher emo- Despite substantial evidence, we still know very little about the
tion literature, studies addressing this relationship are sparse, relationship between EL and EE when it comes to teachers’ emo-
even more so when considering discrete teacher emotions in con- tional experiences during instruction at the state-level – that is,
trast to general affectivity (see Chang, 2009). Kunter et al. (2011) the level of an individual’s actual experiences at the moment. Yet,
found teacher enthusiasm (regarded as a highly positive affective teaching and interacting with students is arguably the most impor-
characteristic of teachers) to be negatively related to burnout (r = – tant task teachers engage in during the course of their professional
0.74 for teaching-related enthusiasm). Similarly, Carson (2006) lives. Furthermore, display rules primarily stem from beliefs about
showed higher levels of teacher burnout corresponded to less pos- what is appropriate around students; thus, we can assume teachers
itive and increased negative emotions. Chang (2013) investigated largely regulate their emotions in class and that emotion regula-
teachers’ episodic emotional experiences and how they relate to tion plays a subordinate role when preparing lessons at home, for
appraisals, different coping strategies, and ultimately to burnout, example. Therefore, an investigation is needed that addresses the
finding a clear relationship between teachers’ burnout rates and the emotional experiences and EL in the context of teacher burnout,
increased intensity of negative emotions from disruptive episodes specifically during instructional activities.
in the classroom. Adding to this evidence, the present study aims to
deepen our understanding of the relationship between EE and dis- ASSESSMENT OF TEACHERS’ EMOTIONAL PROCESSES
crete positive and negative emotional experiences and investigate The majority of studies on teacher emotions and EL employs
them on intra- as well as inter-individual levels. teachers’ generalized self-reports (traits assessed via questionnaires
or interviews). However, theoretical considerations (Robinson
THE RELATIONSHIP BETWEEN TEACHERS’ EMOTIONAL EXHAUSTION and Clore, 2002) and empirical investigations indicate that trait-
AND EMOTIONAL LABOR reports on emotional experiences can be biased and do not
Previous research has identified precursors to teachers’ EE on the necessarily reflect an individual’s actual – state-level – experiences
class- and school-level, such as student misbehavior (Chang and (trait-state discrepancy in teachers’ emotion self-reports; see Keller
Davis, 2009) or school climate (Grayson and Alvarez, 2008), as well et al., 2014). Also, studies employing teachers’ trait-reports only
as on the individual-level, such as self-efficacy (Dicke et al., 2014). address inter-individual differences in teachers’ emotional expe-
Beyond such precursors, EL has been recognized as a central factor riences, and not much is known about how the experiences of
involved in the emergence of EE (e.g., Näring et al., 2006; Judge emotions and EL are related on an intra-individual level. Further-
et al., 2009). more, pertinent EL theories are based on intra-individual, that
Morris and Feldman (1996, p. 987) define EL as “the effort, is, situation-specific considerations, such as in a teaching situation
planning, and control needed to express organizationally desired when teachers experience an inappropriate emotion, they suppress
emotion during interpersonal transactions” and evidence suggests that emotion, thereby draining their resources. That these rela-
that EL is something teachers report to engage in, on a regular basis tionships also extend to the inter-individual level, that is, between
(see for example Sutton, 2004; Meyer, 2009) due to the display rules teachers, is implicitly assumed, yet this may not necessarily be the
in the classroom. Teachers have implicit rules about whether or case.
not, and when and how to display emotions during instructional Some research outside of the teaching profession has identi-
time (Sutton, 2004; Schutz et al., 2007), such as the need to show fied the relationship of state-level emotion labor with EE. Judge
enthusiasm or to remain calm even when class is disrupted. Con- et al. (2009) investigated customer service employees’ state-level
sequently, teachers feel the urge to regulate their emotions, thereby emotions (i.e., emotions are assessed directly at the moment when
engaging in EL. The pertinent, albeit dysfunctional, emotion reg- they are experienced via the ESM (Scollon et al., 2009). Judge et al.
ulation strategy in the context of teacher burnout and particularly (2009) found the degree to which individuals engaged in surface
regarding EE is surface acting (Näring et al., 2006; Chang, 2013). acting on a daily basis was related to their EE. They concluded
Surface acting refers to either suppressing the actual yet unde- that “emotional labor is a dynamic process, wherein the use and
sired emotion (e.g., anger), or faking a desired emotion in order consequences of emotional labor vary between-individuals and
to keep up the idealized image (e.g., Brotheridge and Lee, 2003; within-individuals” (p. 78).
Hochschild, 2012). Carson (2006, Study 2) pioneered the research on teachers’
Research shows that the continuous effort of EL is a stressor on state-level emotions, assessed directly at the moment they are expe-
teachers that draws on their regulatory resources (Muraven et al., rienced, and emotion regulation, using ESM. Teachers were asked
1998; Muraven and Baumeister, 2000) and causes psychological to report their state-level emotions at different times in a day of
strain (e.g., Cheung and Tang, 2007; Diestel and Schmidt, 2011). teaching (e.g., in the mornings or during mid-day breaks), out-
Carson (2007) found surface acting, that is, suppressing, faking, side of instructional time in class. Findings indicated that teacher
or hiding true emotions, led to greater overall burnout for teach- burnout is related to teachers’ emotional experiences, as well as
ers. Tsouloupas et al. (2010) found direct effects between teachers’ the frequency with which teachers regulate their emotions.

Frontiers in Psychology | Educational Psychology December 2014 | Volume 5 | Article 1442 | 2

Keller et al. Teachers’ emotional labor during class

Although much published research focuses on teacher burnout

and its diverse antecedents and consequences, little is known about
state-level emotional processes (such as emotional experiences and
EL) involved in teacher burnout. Even less is known about teachers’
EE based on data drawn from the instructional time or situation,
or at an intra-individual level, from lesson to lesson.

RESEARCH OBJECTIVES AND HYPOTHESES

In response to the notable lack of research addressing teachers’
state-level emotional processes within the context of burnout, this
study aims to investigate the relevance of teachers’ state-level emo-
tional experiences (enjoyment, anxiety, and anger), EL, and how
the occurrence of emotions and EL in actual classroom situations
relates to EE. Particularly, we were initially – on a more exploratory
level – interested in determining how pronounced EL gets for
teachers during instruction time. This was done by drawing on
momentarily assessed EL (state). To our knowledge, ours is the first
study that attempts to assess EL and emotional experiences in vivo,
FIGURE 1 | Figural representation of the study hypotheses. Relations
while teachers are in class. In addition, we investigated how teach- between state-reported variables were investigated both on an inter- and
ers’ trait-reported EE is related to their state reports of emotional intraindividual level. The dashed lines represent the testing of Hypothesis 3
experiences. We therefore formulated our first hypothesis: in which a multilevel regression model was used to test the influences of
trait-reported emotional labor and emotional exhaustion and state-reported
H1: Emotional exhaustion is negatively related to positive emotional experiences on state-reported emotional labor.
emotions (enjoyment) and positively related to negative
emotions (anxiety, anger).
EL, and EE), and handheld devices (Palm Pilot Z22) for the state-
We also expect teachers’ emotional experiences in turn should
based experience sampling assessment. They were instructed to
relate to their EL. We therefore formulated our second hypothe-
fill out the trait questionnaire in advance and then they used
sis:
the handheld device to report state-level data for two consecutive
H2: Enjoyment is negatively related to EL, while anxiety and weeks. Teachers were given a demonstration on how to operate
anger are positively related to EL. the Palm Pilots and were also equipped with a detailed instruc-
tion manual. Questionnaires and handheld devices were collected
Lastly, we investigated how state emotional experiences, trait
3–4 weeks later, since teachers did not all start on the same day.
EE, and trait EL jointly relate to state EL on an intra- and
The handheld devices were programmed with experience sam-
inter-individual level and formulated our third hypothesis:
pling software (PMAT; see Weiss et al., 2004), and data assessment
H3: Emotional exhaustion, trait EL and negative emotions are combined event and random sampling. Teachers were instructed
positively related to state EL, whereas positive emotions are to activate the device at the beginning of each regular lesson (i.e.,
negatively related to state EL. event-sampling). The device was programmed to randomly signal
The relationships pertaining to these hypotheses are depicted in (i.e., random sampling) once within that lesson and presented a
Figure 1. short questionnaire. A 5-min response window was programmed
into the PDA, so that teachers were not forced to interrupt their
MATERIALS AND METHODS lesson in the middle of a sentence. If the teacher did not answer
SAMPLE the question set within 5 min after the initial alarm, the PDA auto-
The participants in the ESM study were 39 teachers (20 female, matically ended the question set and saved it as a “missed signal”;
16 male, 3 did not indicate their gender) from the highest track this happened for about 10% of signals (mostly due to the sig-
of the German school system, the Gymnasium, which approx- naling noise being too low). Verbal feedback given by the teachers
imately one third of students attend (Federal Statistical Office after the ESM-period indicated that teachers were able to imple-
[Statistisches Bundesamt], 2014). The participants were on aver- ment the ESM with relative ease into their teaching and overall,
age 44.14 years old (SD = 11.33 years) and had been teaching they did not find it intrusive. Teachers activated their devices in
for an average of 16.16 years (SD = 11.94 years), including the 20 school lessons on average, and it took them approximately half
induction phase. a minute (M = 37 s, SD = 21 s) to answer the set of questions.
Altogether, the experience sampling assessment yielded N = 794
PROCEDURE state assessments.
After introducing the research project in participating schools’staff
meetings, appointments were made with all interested teachers so MEASURES
they could obtain more information about procedures and tech- Trait-reported emotional exhaustion
nical issues. The teachers were equipped with paper-and-pencil Teachers’ EE was assessed using the respective subscale of the
questionnaires for the trait-based assessment (demographics, trait Maslach Burnout Inventory (Maslach et al., 1996) which was

www.frontiersin.org December 2014 | Volume 5 | Article 1442 | 3

Keller et al. Teachers’ emotional labor during class

translated into German by Enzmann and Kleiber (1989). It con- Regarding Hypotheses 2 and 3, we ran three successive random
sists of nine items which were rated on a five-point scale from intercept and slope models. In Model 1 (M1), state EL is predicted
1 (not true at all) to 5 (completely true), with a sample item by emotional experiences both introduced as groupmean-centered
being “I feel fatigued when I get up in the morning and have Level 1 predictors indicating the within-person effects of emotions
to face another day at school.” The scale showed good reliability on EL, and aggregated grandmean-centered Level 2 predictors
(Cronbach’s Alpha = 0.87). indicating the between-person effects of emotions on EL. In Model
2 (M2), EE and trait EL (as between-level variables) predict
Trait-reported emotional labor state EL. Finally, in Model 3 (M3), all predictors are simul-
To assess teachers’ EL, a modified measure of the Frankfurt Emo- taneously included. For this final model, the equations are as
tion Work Scale (Zapf et al., 1999) by Neubach and Schmidt (2006) follows:
was utilized. Its five items were adapted to match the target group
of teachers (substituting “work” with “class” and “customers” with Level 1 : ELij = β0j + β1j Enjij + β2j Anx ij + β3j Angij + rij ;
“students”). A sample item was as follows: “How often do you have
to show feelings in class that you do not really feel?” The items were Level 2 : β0j = γ00 + γ01 Enjj + γ02 Anx j + γ03 Angj + γ04 EEj
rated on a five-point scale from 1 (never) to 5 (very often). The scale + γ05 ELtraitj + u0j ;
achieved high reliability (Cronbach’s Alpha = 0.91).
β1j = γ10 + u1j ;
State-reported emotional labor β2j = γ20 + u2j ;
To measure teachers’ state-reported EL, two items from the trait EL
scale were adapted to suit the momentary assessment. The items β3j = γ30 + u3j .
were as follows: “At the moment I have to suppress my feelings”and
“At the moment, I have to display emotions that do not correspond We recognize that the influences of the emotions and emotion-
to my inner feelings,” both of which could be rated on a five-point related variables could go both ways in real-life situations.
scale ranging from 1 (not true at all) to 5 (completely true). The However, given the limits in the methodology we have chosen
items represent the two surface acting strategies, namely suppres- (administering survey before collecting ESM data), we did not
sion and faking, and were highly correlated (r = 0.63, p < 0.001 further test for directionality and the reverse effects among these
for within-level and r = 0.86, p < 0.001 for between-level). Both variables. Limitations due to the decisions for the analyses are
items were subsequently combined into on overall scale for further further discussed in a later section.
analysis.
ETHICAL PRINCIPLES
State-reported emotions The present study was conducted abiding by the ethical principles
To represent teachers’ relevant emotional experiences, we chose provided by the German Psychological Society [DGPs] (2007) and
the most frequently experienced positive and negative emotions, the American Psychological Association [APA] (2010). Guidelines
enjoyment and anger, respectively (Keller et al., 2014). Further- provided by these institutions state that formal informed consent is
more, we included anxiety as a particularly detrimental teacher not necessary when no potential harm or distress is to be expected
emotion (see Frenzel et al., 2009; Frenzel, 2014). Due to time and/or when normal educational practices are followed as a goal
constraints for the ESM assessment, we relied on single items to of the research. Prior to their participation, the participants of
assess teachers’ state emotional experiences (for a similar single- the present research were informed of the research, duration, and
item assessment of emotions, see for example, Nett et al., 2011; procedures. Participation was voluntary and participants provided
Goetz et al., 2013). The respective items were formulated as verbal informed consent prior to data collection. All data was
follows: “At the moment, how strongly do you experience enjoy- collected and analyzed anonymously.
ment/anger/anxiety?,” and they could be rated on a five-point scale
from 1 (not at all) to 5 (very strongly). RESULTS
DESCRIPTIVE STATISTICS
STATISTICAL ANALYSES The descriptive statistics of the study variables are given in Table 1.
Our sample represented a nested data structure with measures on As can be seen, the teachers in the current sample reported an
level 1 (N 1 = 794) nested within persons on level 2 (N 2 = 39). average EE of M = 2.27 (SD = 0.66). Enjoyment is the most
To correctly estimate the standard errors in such a nested data prominent emotion teachers reported experiencing while teach-
structure, we conducted multilevel regression analyses (random ing in 99% of the lessons, at least to some extent (rated 2 or
intercepts and slopes) to test for our research hypotheses, using higher; M = 2.81, SD = 0.54), whereas they reported experienc-
the Mplus 7.0 software (Muthén and Muthén, 1998–2012). As ing anxiety only to a very small extent (M = 1.09, SD = 0.49);
such, relationships could be modeled separately for the within- however, anxiety cannot be neglected completely, as teachers indi-
and between-levels. cated feeling anxious at least to some extent in 8% of the lessons
To test Hypothesis 1, we ran three random intercept models (for similar results, see Frenzel et al., 2009). Teachers experienced
with EE (as a between-level variable) predicting the emotional anger at least to some extent in 39% of the lessons (M = 1.61,
experiences of enjoyment, anxiety, and anger (as within-level vari- SD = 0.49). As indicated by the intra-class correlations [ICCs(1)],
ables). The respective equations for these models were Level 1: emotional experiences appear to be quite situation-specific: only
Emotionij = β0j + rij , and Level 2: β0j = γ00 + γ01 EEj + u0j . about 20% of variance in emotions lies between teachers, with the

Frontiers in Psychology | Educational Psychology December 2014 | Volume 5 | Article 1442 | 4

Keller et al. Teachers’ emotional labor during class

Table 1 | Descriptive statistics of study variables. will be; no significant relationship was found with aggregated
state-anxiety. Aggregated state EL is significantly related to
M SD %1 ICC(1) aggregated state-anxiety and anger, yet not to enjoyment. Emo-
tions are not correlated significantly to each other on the
Trait
between-level, meaning that teachers reporting having experi-
Emotional exhaustion 2.27 0.66 – –
enced enjoyment often do not also report less anxiety or anger.
Emotional labor 2.33 0.80 – – Regarding within-level relations, state emotions are correlated
State significantly to each other. Thus, in teaching situations where
Emotional labor 1.48 0.54 38/282 0.39 teachers report having experienced some enjoyment, they also
Enjoyment 2.81 0.54 99 0.23 report less anger and anxiety. Finally, EL is negatively related
to enjoyment and positively related to anxiety and anger on the
Anxiety 1.09 0.18 8 0.19
within-level.
Anger 1.61 0.49 39 0.20

All variables were assessed on a rating scale ranging from 1 to 5. RELATING EMOTIONAL EXHAUSTION TO TEACHERS’ STATE
1 Gives the percentage of lessons, where the respective construct is rated 2 or EMOTIONAL EXPERIENCES
higher. According to previous empirical evidence, teachers’ EE should
2The first value indicates the percentage of lessons for the suppression item, the
second for the faking item.
be mirrored in the experience of diminished levels of posi-
tive and elevated levels of negative emotions while teaching.
largest amount of variance being within teachers, in other words, To address this, we regressed teachers’ state emotional experi-
on the lesson-level. ences (enjoyment, anxiety, and anger) on their trait reported
Teachers engage to a moderate extent in EL (trait: M = 2.33, EE (Hypothesis 1; see Table 3). As hypothesized, EE relates
SD = 0.80; state: M = 1.48, SD = 0.54). In the momentary assess- negatively to teachers’ enjoyment (b = –0.35, p < 0.05) and
ments, altogether with the 794 responses, teachers indicate that positively to teacher anger (b = 0.25, p < 0.01). The effect
they suppress their emotions more often (in 38% of the lessons) sizes are moderate (R2 enjoyment/anger = 0.20/0.16). There was no
than they fake emotions (28%). When compared to emotional significant relationship between EE and teachers’ experiences of
experiences, EL seems to be more person-specific; 39% of the anxiety.
variance is between teachers, yet the largest amount is still within
teachers on the lesson-level. RELATING EMOTIONAL EXPERIENCES AND EMOTIONAL EXHAUSTION
Intercorrelations of all study variables are given in Table 2. TO STATE EMOTIONAL LABOR
On the between-level, EE is positively related to trait EL, but In order to untangle the relations of teachers’ emotional
not to (aggregated) state EL (only having a marginally signifi- experiences and EE and their state-reported EL (Hypothe-
cant relationship, p < 0.10). Furthermore, the more exhausted ses 2 and 3), we ran three successive regression models (see
teachers indicate they were, the lower their aggregated state Table 4).
enjoyment, and the higher the reported aggregated state anger On the within-level (see M1), all emotions are predictive for
EL as hypothesized, with enjoyment being negatively related to EL
(b = –0.10, p < 0.01) and anxiety and anger positively related to
Table 2 | Intercorrelations of study variables. EL (banxiety = 0.30, p < 0.01, banger = 0.34, p < 0.001). In turn, on
the between-level, only anger is predictive of state EL (b = 0.88,
(1) (2) (3) (4) (5) (6) p < 0.001). In total, 39 and 67% of variance in state EL are
Trait
(1) Emotional 1 0.51*** 0.22 −0.44** 0.14 0.40**
Table 3 | Predicting teachers’ state emotional experiences by trait
exhaustion reported emotional exhaustion.
(2) Emotional − 1 0.48*** −0.44** 0.33 0.39**
labor Enjoyment Anxiety Anger

State Estimate SE Estimate SE Estimate SE

(3) Emotional − − 1 −0.08 0.51* 0.90***
Intercept (γ00 ) 2.76 0.08 1.09 0.03 1.61 0.07
labor
Emotional –0.35* 0.15 0.03 0.04 0.25** 0.08
(4) Enjoyment − − −0.35*** 1 −0.04 −0.14
exhaustion (γ01 )
(5) Anxiety − − 0.24*** −0.18*** 1 0.44
R2 0.20 0.02 0.16
(6) Anger − − 0.48*** −0.42*** 0.13* 1
Unstandardized coefficients are shown. Emotional exhaustion was entered as a
The values below the diagonal give the intercorrelations on the within-level, above grandmean-centered predictor into the respective model. The explained variance
the diagonal for the between level. R2 refers to the explained variance on the between level.
*p < 0.05, **p < 0.01, ***p < 0.001. *p < 0.05, **p < 0.01.

www.frontiersin.org December 2014 | Volume 5 | Article 1442 | 5

Keller et al. Teachers’ emotional labor during class

Table 4 | Predicting teachers’ state emotional labor by emotional experiences, emotional exhaustion, and trait emotional labor.

Emotional labor (state)

M1 M2 M3

Estimate SE Estimate SE Estimate SE

Within
Enjoyment (γ10 ) −0.10** 0.03 −0.11** 0.03
Anxiety (γ20 ) 0.30** 0.11 0.30** 0.11
Anger (γ30 ) 0.34*** 0.05 0.34*** 0.05
Slope variance
Enjoyment (Var u 1j ) 0.01 0.01 0.01 0.01
Anxiety (Var u 2j ) 0.06 0.06 0.03 0.03
Anger (Var u 3j ) 0.06*** 0.01 0.03* 0.01
Between
Enjoyment (γ01 ) 0.02 0.09 0.07 0.07
Anxiety (γ02 ) 0.63 0.36 0.43 0.30
Anger (γ03 ) 0.88*** 0.21 0.89*** 0.20
Emotional exhaustion (γ04 ) −0.02 0.10 −0.15 0.09
Emotional labor (trait) (γ05 ) 0.35** 0.12 0.22*** 0.06
R2
Within 0.39 0.38
Between 0.67 0.24 0.73

Unstandardized coefficients are shown. Level 1-predictors are entered groupmean centered into the model. The respective emotions as predictors on level 2 were
grandmean centered and introduced as the per-person aggregated emotions based on the level 1 data.
*p < 0.05, **p < 0.01, ***p < 0.001.

explained by emotional experiences on the within- and between- THE ROLE OF EMOTIONAL EXHAUSTION FOR TEACHERS’
levels, respectively. Interestingly, the slope variance for anger is MOMENTARILY EXPERIENCED EMOTIONS
small, yet statistically significant, indicating that the relationship Regression analyses showed that teachers’ overall level of EE is
between momentarily experienced anger and EL differs between indeed reflected in their emotional experiences while in class: the
teachers. more exhausted teachers indicated they were, the less experiences
The relationship between EE and state-reported EL is close of enjoyment and the more experiences of anger they indicated.
to zero when controlling for trait EL (M2). Both trait variables The experience of anxiety was not related to EE. Only a handful
explain 24% of variance in teachers’ state EL. When comparing of studies have directly examined this relationship, and our find-
explained variances on the between-level in M1, M2, and M3, it ings confirm the previously shown relationship between negative
can be seen that 6% of explained variance is unique to trait EL and emotions and EE (Carson, 2007; Chang, 2013).
trait EE. The majority of previous studies on teacher burnout do
Introducing all predictors into the regression equation (M3) not examine teachers’ (discrete) emotions related to teaching in
leaves the coefficients for emotional experiences fairly unchanged: class as either consequences of or antecedents (or both) to EE
on the within-level, all emotions are predictive of state EL; on the as the core dimension of burnout. In one recent correlational
between-level, only anger and trait EL are statistically significant study, Chang (2013) tested both directions of the relationship
as predictors of state EL. (burnout leads to negative emotions or negative emotions leads
to burnout) and concluded that the intensity of negative emo-
DISCUSSION tions from one episode accounted for teacher burnout. Our
The purpose of the study was to investigate teachers’ emotional study adds new understandings of such a relationship by con-
processes, or their experiences of discrete emotions and EL, and firming that EE (trait-level) could contribute to the experience
relate it to EE. The ESM was employed to tackle these emotional of enjoyment and anger (state-level); we didnot test for the
processes on a state level and assess them during in-class instruc- reverse effect (emotional experiences on EE) because EE was
tion, thus allowing for intraindividual analyses. This is the first assessed prior to the ESM-period. Given the mixed findings of
study to do so. the directions, we believe future studies could continue unpacking

Frontiers in Psychology | Educational Psychology December 2014 | Volume 5 | Article 1442 | 6

Keller et al. Teachers’ emotional labor during class

this complex and dynamic relationship by employing a qualita- Given that EL is clearly associated with EE, identifying the
tive approach (by interviewing teachers, for example) coupled factors that lead some teachers to deal with their anger in a
with a quantitative design that allows for determining causal given teaching situation more adaptively might be beneficial to
links. prevent exhaustion.

THE RELATION OF TEACHERS’ MOMENTARILY EXPERIENCED METHODOLOGICAL CONSIDERATIONS ON THE ASSESSMENT OF

EMOTIONS TO EMOTIONAL LABOR TEACHERS’ EMOTIONAL PROCESSES
Analyses revealed that teachers regularly suppress or fake their To overcome several drawbacks from previous studies on teach-
emotions. The teachers in our sample reported having engaged ers’ emotional lives, we employed the ESM (Carson et al., 2010;
in EL and employing surface acting strategies (suppression or Keller et al., 2014) to assess teachers’ emotional experiences and
faking) to regulate their state emotions in about one third of EL momentarily while they were in class and teaching. In partic-
the covered lessons. In addition, EL is significantly related to ular, trait emotions, that is, emotions as assessed on a generalized
teachers’ experiences of anger on an inter-individual level. This level, are assumed to be biased and do not necessarily reflect
is congruent with previous results from inter-individual analyses the actual emotions as experienced in the situation (see Robin-
that indicated the prevalent relevance of anger within the con- son and Clore, 2002). Thus, the present assessment of teachers’
text of EL and EE (Sutton and Wheatley, 2003; Chang and Davis, state emotions instead of commonly used trait emotions can over-
2009). Teachers’ experience of anxiety is only related to EL on an come this methodological flaw and provide insight into teachers’
intra-individual level, yet no such relation exists when comparing momentary emotional experiences in a highly ecologically valid
teachers. way.
Regarding teachers’ experience of enjoyment, a negative rela- Related to this, we found EL as assessed during a concrete
tionship between enjoyment as the desired emotional experience teaching situation to be only moderately related to EL as assessed
(as indicated by implicit display rules) and EL was anticipated. on a generalized trait-level. The question arises regarding how
Intra-individually, this was supported by the study results: a reliably trait reports capture teachers’ emotion-related constructs
teacher who experiences more enjoyment in a given teaching situ- as they actually occur in a given teaching situation. This issue
ation reports lower levels of EL. This conclusion cannot be drawn should be addressed and explicitly investigated in future stud-
based on inter-individual results; here, the relation between enjoy- ies. We also found anxiety to be only of subordinate importance
ment and EL is close to zero. One explanation for this finding when it came to teachers’ emotional lives in class. Two reasons
may be the differences of interrelations between emotions on the seem likely to explain this finding. First, the overall low values
intra- and inter-individual levels: whereas emotional experiences of state-anxiety could be due to employing the actual emotion-
correspond to each other on a situational basis (state correla- word for assessing the respective emotion. The word “anxiety”
tions; see Table 2), teachers who experience more enjoyment do implies high arousal levels, which presumably occur very seldom
not necessarily also report less anger (lack of inter-correlations during a lesson and would be captured by a random assessment
on an inter-individual level). Thus, in a given teaching situ- even less often; yet, anxiety as an emotion covers also low-arousal
ation, the experience of enjoyment goes along with reduced states of anxiety, such as nervousness, which might not be cov-
levels of anxiety and anger, thus also reducing the necessity ered by our assessment (compared with somewhat higher values
for engaging in EL. Lacking the inter-individual relationships for teachers’ state-anxiety using the item wording: “I was tense
between emotions, this explanation cannot hold true for the and nervous during this lesson” in Frenzel et al., 2009). Second,
(lack of) inter-individual relationship between enjoyment and and related to the previously mentioned trait-state ambiguity, is
EL. Future studies could address this issue and investigate the the fact that trait emotions might reflect something other than
relations between emotions as they are intra-individually expe- the actual emotions experienced in a concrete situation. Thus,
rienced in situations, and to what extent these relations hold teachers’ trait-reported emotions differ in their magnitude from
true for inter-individual differences. Also, future investigations state-reported emotions, including anxiety (Keller et al., 2014).
could address the specific role enjoyment plays in EL and emotion As such, the present findings of low anxiety levels while teaching
regulation in general, by, for example, including other emotion might be a first indicator that anxiety occurs less frequently in the
regulation strategies than those considered within the present actual teaching situation, but more so in retrospect when evalu-
study. ating situations over a longer time frame and employing personal
Regarding the strength of the relationship between teachers’ beliefs when doing so.
momentarily experienced emotions and EL, there seem to be Teachers spend the majority of their time teaching, and this can
differences between teachers regarding the emotion of anger; be considered the central task in which they engage (OECD, 2011);
in other words, how strongly anger relates to EL on a situa- yet their emotional lives in the classroom have not been explicitly
tional level differs between teachers. This could be indicative addressed to date. Thus, the ESM allows us to tap into teachers’
that, given a specific level of anger, some teachers regulate their emotional lives during instruction and covers an important —
anger expression to a larger extent than do other teachers. This perhaps the most important — aspect of teachers’ professional
might be due to teacher and/or situational differences in emotion lives. While in class, complex interactions with students require
regulation strategies not covered within the present investiga- teachers to constantly monitor and regulate their affective image,
tion (e.g., deep acting; Brotheridge and Lee, 2003; Hochschild, thereby drawing on self-regulatory resources. Thus, addressing
2012) and would warrant further investigations in the future. emotions and EL while teaching is highly relevant.

www.frontiersin.org December 2014 | Volume 5 | Article 1442 | 7

Keller et al. Teachers’ emotional labor during class

Lastly, the ESM assessment as utilized in the present study investigation shows that experiences of anger necessitate teach-
allows for intra-individual analyses of teachers’ emotional pro- ers to engage in EL and are also a correlate of teachers’ EE. One
cesses. Unraveling intra-individual functioning is a core goal in may speculate that exhaustion leads to more frequent experiences
personality psychology (Eid and Diener, 1999); yet, the majority of anger, or exhaustion is a consequence of increased experi-
of research on teachers’ emotions focuses on inter-individual dif- ences of anger, or both. Thus, findings of the present study
ferences. In the present study, we were able to separate differences imply the beneficial effects of anger reduction, which should
between teachers from differences occurring across situations, yet lead to less EL, and possibly over a longer time frame, also to
within-teachers. As findings on the relationship between emotions less exhaustion. Reducing one’s experience of anger might be
and EL indicate, the results that were yielded on an intra-individual viable by so-called reappraisal strategies (see for example Gross
level are not necessarily transferable to the inter-individual level. and John, 1998). Future studies could develop intervention pro-
grams for teachers based on reappraisal training for a reduction
LIMITATIONS of anger experiences during class and investigate the effects of
Due to the ESM design, the present investigation is subject to such training on EL and consequently, exhaustion or well-being
some limitations. First, all variables as implemented in the current in teachers.
study were assessed via self-reports. While that may be justified, Beyond the prevalent importance of anger, the present study’s
as only the individuals themselves can report on their concrete results also indicate the potential of enjoyment in reducing teach-
subjective affective experiences and stress-related variables, other ers’ EL: on a situational level, increasing the experience of positive
measures complementing self-reports, such as physiological mea- emotions might decrease teachers’ engagement in EL, thus reduc-
sures of arousal, are called for and could be implemented in future ing their risk of eventually suffering from exhaustion and burnout.
studies. Positive emotions have previously been suggested to act as an
The teacher sample in the present study is rather small. The important resource (Fredrickson, 1998), yet empirical results
ESM assessment yielded an adequate sample size on the within- backing that theoretical claim have been missing so far. Beyond the
level, however, multilevel analyses demand between 30 and 50 implication that enjoyment as an appropriate emotion (according
units on the between-level for reliably estimating between-level to implicit display rules) demands less EL efforts, the negative rela-
effects and differences (Maas and Hox, 2004). Thus, the present tionship between enjoyment and EE on an intra-individual level
sample of 39 teachers should yield reliable results, yet a replication points toward its importance as a possible resource in the teach-
of study findings would be helpful. Also, the teachers participating ing context. Thus, increasing teachers’ experiences of enjoyment
in the present study all teach in one school track (the Gymnasium could act as a buffer for teacher burnout.
in Germany); future studies could also consider other school types Although our study hints at this possible relationship and its
to gain a more comprehensive picture. implications, further efforts are needed to explicitly investigate
Strictly speaking, our cross-sectional design does not allow this link. Thus, future investigations should focus on emotional
us to model causal effects. In fact, regarding emotional pro- experiences of teachers and how they — causally — relate to EL
cesses in the context of teachers’ stress and EE, effects are most and burnout on an intra-individual level, including the charac-
likely reciprocal. For example, given that EE leads to an increase teristics of the situation and how it is perceived and appraised by
in the experience of negative emotions, these in turn would teachers (on the role of appraisals and coping strategies within
necessitate an increase in EL efforts, thereby depleting resources the context of burnout, see for example, Chang, 2009, 2013).
that could ultimately lead to higher levels of exhaustion. Future Including situation characteristics and how they are perceived and
studies could combine momentary assessment with a longitudi- appraised by teachers would allow for identifying teaching situ-
nal study design to unravel both intra-individual and long-term ations (as characterized by student behavior, for example) and
processes and effects. how they influence teachers’ emotional experiences, which would
ultimately provide the means for interventions designed to shape
CONCLUSION beneficial emotional experiences that reduce teachers’ risk for
The present study fills a gap in the existing literature by inves- burnout.
tigating emotional processes within the context of teachers’ EE
(see Chang, 2009). It employs the ESM to assess teachers’ discrete
REFERENCES
momentary emotions and EL while teaching. This approach allows Abraham, R. (1999). Emotional dissonance in organizations: conceptualizing the
us to address intra-individual emotional processes and relate them roles of self-esteem and job-induced tension. Leadersh. Organ. Dev. J. 20, 18–25.
to inter-individual differences in teachers’ EE. doi: 10.1108/01437739910251152
In investigating emotional processes as they occur in a given American Psychological Association [APA]. (2010). Publication Manual of the Amer-
ican Psychological Association, 6th Edn. Washington, DC: American Psychological
teaching situation, the present study cannot draw any conclusions Association.
as to what causes teachers’ emotional reactions in class. Previous Ashforth, B. E., and Lee, R. T. (1990). Defensive behavior in organizations: a
investigations have shown the paramount importance of students’ preliminary model. Hum. Relat. 43, 621–648. doi: 10.1177/001872679004300702
(mis)behavior in the context of teacher EL and exhaustion (Chang Brotheridge, C. M., and Grandey, A. A. (2002). Emotional labor and burnout:
and Davis, 2009; Tsouloupas et al., 2010). Thus, students’ lack comparing two perspectives of “people work”. J. Vocat. Behav. 60, 17–39. doi:
10.1006/jvbe.2001.1815
of adherence to stated classroom rules or obstruction of teach- Brotheridge, C. M., and Lee, R. T. (2003). Development and validation of
ers’ goals could cause teachers to experience anger (Sutton and the emotional labour scale. J. Occup. Organ. Psychol. 76, 365–379. doi:
Wheatley, 2003; Sutton, 2007; Frenzel et al., 2009); the present 10.1348/096317903769647229

Frontiers in Psychology | Educational Psychology December 2014 | Volume 5 | Article 1442 | 8

Keller et al. Teachers’ emotional labor during class

Brouwers, A., and Tomic, W. (2000). A longitudinal study of teacher burnout and Hakanen, J. J., Bakker, A. B., and Schaufeli, W. B. (2006). Burnout and work
perceived self-efficacy in classroom management. Teach. Teach. Educ. 16, 239– engagement among teachers. J. Sch. Psychol. 43, 495–513. doi: 10.1016/j.jsp.2005.
253. doi: 10.1016/S0742-051X(99)00057-8 11.001
Carson, R. L. (2006). Exploring the Episodic Nature of Teachers’ Emotions as it Relates Hargreaves, A. (1998). The emotional practice of teaching. Teach. Teach. Educ. 14,
to Teacher Burnout. West Lafayette: Purdue University. 835–854. doi: 10.1016/s0742-051x(98)00025-0
Carson, R. L. (2007). Emotional regulation and teacher burnout: who says that Hochschild, A. R. (2012). The Managed Heart: Commercialization of Human Feeling.
the management of emotional expression doesn’t matter? Paper presented at the Berkeley, CA: University of California Press.
American Educational Research Association Annual Meeting, Chicago, IL. Judge, T. A., Woolf, E. F., and Hurst, C. (2009). Is emotional labor more difficult for
Carson, R. L., Weiss, H. M., and Templin, T. J. (2010). Ecological momentary some than for others? A multilevel, experience-sampling study. Pers. Psychol. 62,
assessment: a research method for studying the daily lives of teachers. Int. J. Res. 57–88. doi: 10.1111/j.1744-6570.2008.01129.x
Method Educ. 33, 165–182. doi: 10.1080/1743727X.2010.484548 Keller, M. M., Frenzel, A. C., Goetz, T., Pekrun, R., and Hensley, L. (2014). “Exploring
Chang, M.-L. (2009). An appraisal perspective of teacher burnout: examining the teacher emotions: a literature review and an experience sampling study,” in Teacher
emotional work of teachers. Educ. Psychol. Rev. 21, 193–218. doi: 10.1007/s10648- Motivation: Theory and Practice, eds P. W. Richardson, S. A. Karabenick, and H.
009-9106-y M. G. Watt (New York: Routledge), 69–82.
Chang, M.-L. (2013). Toward a theoretical model to understand teacher emotions Klusmann, U., Kunter, M., Trautwein, U., Lüdtke, O., and Baumert, J. (2008).
and teacher burnout in the context of student misbehavior: appraisal, regulation, Teachers’ occupational well-being and quality of instruction: the important role
and coping. Motiv. Emot. 37, 799–817. doi: 10.1007/s11031-012-9335-0 of self-regulatory patterns. J. Educ. Psychol. 100, 702–715. doi: 10.1037/0022-
Chang, M.-L., and Davis, H. A. (2009). “Understanding the role of teacher appraisals 0663.100.3.702
in shaping the dynamics of their relationships with students: deconstructing Kunter, M., Frenzel, A. C., Nagy, G., Baumert, J., and Pekrun, R. (2011). Teacher
teachers’ judgement of disruptive behavior/ students,” in Advances in Teacher enthusiasm: dimensionality and context specificity. Contemp. Educ. Psychol. 36,
Emotion Research: The Impact on Teachers’ Lives, eds P. A. Schutz and M. Zembylas 289–301. doi: 10.1016/j.cedpsych.2011.07.001
(Dordrecht: Springer), 95–127. Maas, C. J. M., and Hox, J. J. (2004). Robustness issues in multilevel regression
Cheung, F. Y.-L., and Tang, C. S.-K. (2007). The influence of emotional dissonance analysis. Stat. Neerl. 58, 127–137. doi: 10.1046/j.0039-0402.2003.00252.x
and resources at work on job burnout among chinese human service employees. Macdonald, D. (1999). Teacher attrition: a review of literature. Teach. Teach. Educ.
Int. J. Stress Manag. 14, 72–87. doi: 10.1037/1072-5245.14.1.72 15, 835–848. doi: 10.1016/S0742-051X(99)00031-1
de Heus, P., and Diekstra, R. F. W. (1999). “Do teachers burn out more easily? A Maslach, C., Jackson, S. E., and Schwab, R. L. (1996). “Maslach burnout
comparison of teachers with other social professions on work stress and burnout inventory – educators survey (mbi-es),” in Mbi Manual, 3rd Edn, eds C. Maslach,
symptoms,” in Understanding and Preventing Teacher Burnout: A Sourcebook of S. E. Jackson, and M. P. Leiter (Palo Alto, CA: Consulting Psychologists Press).
International Research and Practice, eds R. Vandenberghe and A. M. Huberman Maslach, C., Schaufeli, W. B., and Leiter, M. P. (2001). Job burnout. Annu. Rev.
(Cambridge: Cambridge University Press), 269–284. Psychol. 52, 397–422. doi: 10.1146/annurev.psych.52.1.397
Dicke, T., Parker, P. D., Marsh, H. W., Kunter, M., Schmeck, A., and Leutner, Meyer, D. K. (2009). “Entering the emotional practices of teaching,” in Advances
D. (2014). Self-efficacy in classroom management, classroom disturbances, and in Teacher Emotion Reserach: The Impact on Teachers’ Lives, eds P. A. Schutz
emotional exhaustion: a moderated mediation analysis of teacher candidates. and M. Zembylas (Dordrecht: Springer), 73–91. doi: 10.1007/978-1-4419-
J. Educ. Psychol. 106, 569–583. doi: 10.1037/a0035504 0564-2_5
Diestel, S., and Schmidt, K.-H. (2011). Costs of simultaneous coping with emotional Morris, J. A., and Feldman, D. C. (1996). The dimensions, antecedents, and
dissonance and self-control demands at work: results from two german samples. consequences of emotional labor. Acad. Manag. Rev. 21, 986–1010. doi:
J. Appl. Psychol. 96, 643–653. doi: 10.1037/a0022134 10.5465/AMR.1996.9704071
Eid, M., and Diener, E. (1999). Intraindividual variability in affect: reliability, Muraven, M., and Baumeister, R. F. (2000). Self-regulation and depletion of limited
validity, and personality correlates. J. Pers. Soc. Psychol. 76, 662–676. doi: resources: does self-control resemble a muscle? Psychol. Bull. 126, 247–259. doi:
10.1037/0022-3514.76.4.662 10.1037/0033-2909.126.2.247
Enzmann, D., and Kleiber, D. (1989). Helfer Leiden. Streß und Burnout in psy- Muraven, M., Tice, D. M., and Baumeister, R. F. (1998). Self-control as a limited
chosozialen Berufen [Helpers’ Suffering: Stress and Burnout in Psychosocial resource: regulatory depletion patterns. J. Pers. Soc. Psychol. 74, 774–789. doi:
Occupations]. Heidelberg: Asanger. 10.1037/0022-3514.74.3.774
Evers, W. J. G., Tomic, W., and Brouwers, A. (2004). Burnout among teachers: Muthén, B. O., and Muthén, L. K. (1998–2012). Mplus User’s Guide, 7th Edn. Los
students’ and teachers’ perceptions compared. Sch. Psychol. Int. 25, 131–148. doi: Angeles, CA: Muthén & Muthén.
10.1177/0143034304043670 Näring, G., Briët, M., and Brouwers, A. (2006). Beyond demand–control: emotional
Federal Statistical Office [Statistisches Bundesamt]. (2014). Schulen auf einen Blick labour and symptoms of burnout in teachers. Work Stress 20, 303–315. doi:
[Schools at a Glance]. Available at: https://www.destatis.de/DE/Publikationen/ 10.1080/02678370601065182
Thematisch/BildungForschungKultur/Schulen/BroschuereSchulenBlick.html Nett, U. E., Goetz, T., and Hall, N. (2011). Coping with boredom in school:
Fredrickson, B. L. (1998). What good are positive emotions? Rev. Gen. Psychol. 2, an experience sampling perspective. Contemp. Educ. Psychol. 36, 49–59. doi:
300–319. doi: 10.1037/1089-2680.2.3.300 10.1016/j.cedpsych.2010.10.003
Frenzel, A. C. (2014). “Teacher emotions,” in Handbook of Emotions in Education, Neubach, B., and Schmidt, K.-H. (2006). Selbstkontrolle als Arbeitsanforderung:
eds L. Linnenbrink-Garcia and R. Pekrun (New York: Routledge), 494–519. Rekonzeptualisierung und Validierung eines Messinstruments [Self-control as
Frenzel, A. C., Goetz, T., Stephens, E. J., and Jacob, B. (2009). “Antecedents and job requirement: Reconceptualization and validation of a measuring instru-
effects of teachers’ emotional experiences: an integrated perspective and empirical ment]. Z. Arbeits Organisationspsychol. 50, 103–109. doi: 10.1026/0932-4089.50.
test,” in Advances in Teacher Emotion Research, eds P. A. Schutz and M. Zembylas 2.103
(Dordrecht: Springer), 129–151. Nias, J. (1996). Thinking about feeling: the emotions in teaching. Camb. J. Educ. 26,
German Psychological Society [DGPs]. (2007). Richtlinien zur Manuskriptgestal- 293–306. doi: 10.1080/0305764960260301
tung. Göttingen: Hogrefe. OECD. (2011). Education at a Glance 2011: OECD Indicators. Paris: OECD
Goetz, T., Lüdtke, O., Nett, U. E., Keller, M., and Lipnevich, A. (2013). Publishing. doi: 10.1787/eag-2011-en
Characteristics of teaching and students’ emotions in the classroom: investi- Robinson, M. D., and Clore, G. L. (2002). Belief and feeling: evidence for an
gating differences across domains. Contemp. Educ. Psychol. 38, 383–394. doi: accessibility model of emotional self-report. Psychol. Bull. 128, 934–960. doi:
10.1016/j.cedpsych.2013.08.001 10.1037/0033-2909.128.6.934
Grayson, J. L., and Alvarez, H. K. (2008). School climate factors relating to Schutz, P., Cross, D. I., Hong, J. Y., and Osbon, J. N. (2007). “Teacher iden-
teacher burnout: a mediator model. Teach. Teach. Educ. 24, 1349–1363. doi: tities, beliefs, and goals related to emotions in the classroom,” in Emotion
10.1016/j.tate.2007.06.005 in education, eds P. A. Schutz and R. Pekrun (Burlington: Academic Press),
Gross, J. J., and John, O. P. (1998). The emerging field of emotion regulation: an 223–241.
integrative review. Rev. Gen. Psychol. 2, 271–299. doi: 10.1037/1089-2680.2.3.271 Schutz, P. A., and Pekrun, R. (2007). Emotion in Education. Amsterdam: Elsevier.

www.frontiersin.org December 2014 | Volume 5 | Article 1442 | 9

Keller et al. Teachers’ emotional labor during class

Scollon, N. C., Kim-Prieto, C., and Diener, E. (2009). Experience sampling: promises Wolpin, J., Burke, R. J., and Greenglass, E. R. (1991). Is job satisfaction an antecedent
and pitfalls, strengths and weaknesses. Soc. Indic. Res. Ser. 39, 157–180. doi: or a consequence of psychological burnout? Hum. Relat. 44, 193–209. doi:
10.1007/978-90-481-2354-4_8 10.1177/001872679104400205
Skaalvik, E. M., and Skaalvik, S. (2007). Dimensions of teacher self-efficacy Zapf, D., Vogt, C., Seifert, C., Mertini, H., and Isic, A. (1999). Emotion work as
and relations with strain factors, perceived collective teacher efficacy, and a source of stress: the concept and development of an instrument. Eur. J. Work
teacher burnout. J. Educ. Psychol. 99, 611–625. doi: 10.1037/0022-0663.99. Organ. Psychol. 8, 371–400. doi: 10.1080/135943299398230
3.611
Sutton, R. E. (2004). Emotional regulation goals and strategies of teachers. Soc. Conflict of Interest Statement: The authors declare that the research was conducted
Psychol. Educ. 7, 379–398. doi: 10.1007/s11218-004-4229-y in the absence of any commercial or financial relationships that could be construed
Sutton, R. E. (2007). “Teachers’ anger, frustration, and self-regulation,” in Emotion as a potential conflict of interest.
in Education, eds P. A. Schutz and P. Reinhard (Burlington: Academic Press),
259–274. Received: 10 September 2014; accepted: 25 November 2014; published online: 11
Sutton, R. E., and Wheatley, K. F. (2003). Teachers’ emotions and teaching: a review December 2014.
of the literature and directions for future research. Educ. Psychol. Rev. 15, 327–358. Citation: Keller MM, Chang M-L, Becker ES, Goetz T and Frenzel AC (2014) Teachers’
doi: 10.1023/A:1026131715856 emotional experiences and exhaustion as predictors of emotional labor in the classroom:
Tsouloupas, C. N., Carson, R. L., Matthews, R., Grawitch, M. J., and an experience sampling study. Front. Psychol. 5:1442. doi: 10.3389/fpsyg.2014.01442
Barber, L. K. (2010). Exploring the association between teachers’ perceived This article was submitted to Educational Psychology, a section of the journal Frontiers
student misbehaviour and emotional exhaustion: the importance of teacher in Psychology.
efficacy beliefs and emotion regulation. Educ. Psychol. 30, 173–189. doi: Copyright © 2014 Keller, Chang, Becker, Goetz and Frenzel. This is an open-access arti-
10.1080/01443410903494460 cle distributed under the terms of the Creative Commons Attribution License (CC BY).
Weiss, H. M., Beal, D. J., Lucy, S. L., and MacDermid, S. M. (2004). Con- The use, distribution or reproduction in other forums is permitted, provided the original
structing Ema Studies with Pmat: The Purdue Momentary Assessment Tool author(s) or licensor are credited and that the original publication in this journal is cited,
User’s Manual. West Lafayette, IN: Military Research Institute at Purdue in accordance with accepted academic practice. No use, distribution or reproduction is
University. permitted which does not comply with these terms.

Frontiers in Psychology | Educational Psychology December 2014 | Volume 5 | Article 1442 | 10