Emotion Work as a Source of Stress: The Concept and Development of an Instrument

EMOTION WORK

371

EUROPEAN JOURNAL OF WORK AND ORGANIZATIONAL PSYCHOLOGY, 1999, 8 (3), 371–400

Requests for reprints should be addressed to D. Zapf, Department of Psychology, J.W. Goethe-

University Frankfurt, Mertonstr. 17, D-60054 Frankfurt, Germany. Email: D.Zapf@psych.uni-

frankfurt.de.

The hotel study was supported by the Berufsgenossenschaft Nahrung. Special thanks are due to

P. Bärenz and A. Landgraf.

Emotion Work as a Source of Stress:

The Concept and Development of an Instrument

Dieter Zapf, Christoph Vogt, Claudia Seifert, Heidrun Mertini,

and Amela Isic

J.W. Goethe-University Frankfurt, Germany

This article discussed emotion work as a neglected area in organizational stress

research. Emotion work (emotional labour) was defined as the emotional

regulation required of the employees in the display of organizationally desired

emotions. Based on the existing literature on emotion work and action theory,

emotional regulation requirements (sub-scales: the requirement to express positive

emotions; the requirement to express and handle negative emotions, the

requirement to be sensitive to clients’ emotions, and the requirement to show

sympathy), emotional regulation possibilities (control), and emotional regulation

problems (emotional dissonance) were differentiated. Questionnaires were

developed and applied in a sample of employees in a handicapped children’s home

(N = 83), in the hotel business (N = 175) and employees working in call-centres

(N = 250). Scales showed satisfactory reliabilities. Exploratory and confirmatory

factor analyses revealed minor problems with discriminant validity of the scales.

Construct validation showed that the emotion work scales were both positively and

negatively related with psychological health.

Psychological stress research in organizations comprises a substantial body of

research spanning the last 20 years, and demonstrating significant relations

between psychological stressors and strains. In 1989 the European Council

passed various laws to improve the protection of workers’ health and promote

measures of health and well-being. In many European countries, these laws still

have to be translated into national law. In Germany, this was achieved in 1996 by

passing a new law for occupational health and safety (“Arbeitsschutzgesetz”).

The new law addressed physical stressors such as carrying heavy weights or one-

sided and unusual body positions, environmental stressors such as noise or

temperature, which have been investigated in the human factors literature, and

372

ZAPF ET AL.

psychological stressors. Although the former were explicitly mentioned, the

aspects these psychological stressors should comprise were not. Most studies on

psychological stressors at work measure stressors that are related to the work

tasks and to the organization of work (cf. Kahn & Byosiere, 1992; Zapf,

Dormann, & Frese, 1996). Typical examples are quantitative and qualitative

overload or time pressure. There are only a few approaches that try to systematize

psychological job stressors based on a general framework. The most prominent

approaches use role theory to link different role demands such as role conflict,

role ambiguity and role overload to psychological stress (Kahn, Wolfe, Quinn,

Snoek, & Rosenthal, 1964). Another approach is the differentiation of job

stressors according to their effect on action regulation (Frese & Zapf, 1994;

Greiner & Leitner, 1989; Semmer, 1984; Semmer, Zapf, & Dunckel, 1995, 1999;

Zapf, 1993). Basically, these stressors are of a cognitive nature, that is, working

conditions are considered to be stressful because they negatively affect various

aspects of information processing during task execution and because they require

mental effort. Examples are time pressure, interruptions, concentration

necessities or uncertainty at work. Another perspective examines psychological

stressors associated with social relations at work. Scales addressing social

stressors (Frese & Zapf, 1987) or interpersonal conflict scales (Spector, 1987)

measure conflicts, animosities, verbal aggression and unjust behaviour at work.

The theories underlying these kinds of stressors typically relate to conflict and

aggression.

Burnout is yet another research area that points to job requirements not

included in the concepts of psychological job stressors mentioned so far. Burnout

was first investigated in the helping professions (Maslach, 1982; Maslach &

Leiter, 1997; Schaufeli & Buunk, 1996; Schaufeli & Enzmann, 1998; Schaufeli,

Maslach, & Marek, 1993). It is argued that the personal relationships with

patients, clients, or children are very demanding and require a high amount of

empathy and emotional involvement. This is usually combined with a high

aspiration level to build up personal relationships and avoid treating other people

like objects. In these professions, the management of emotions is considered a

central part of work. Burnout is then an indication that employees are no longer

able to adequately manage their emotions when interacting with clients. It is a

syndrome consisting of three aspects: emotional exhaustion, depersonalization,

and reduced personal accomplishment (Maslach & Jackson, 1986). It is argued

that, in the long run, burnout leads to psychosomatic complaints, depression, and

other long-term stress effects. A recent meta-analysis (Lee & Ashforth, 1996) of

the existing literature found role stress to be one of the best predictors of burnout

variables. Interestingly, studies on burnout did not try to directly measure the

emotional aspects at work. Rather, these aspects were taken as a given by doing

research with samples where emotional job requirements could be taken for

granted. Instead, various job stressors, such as role conflict, role ambiguity, time

pressure, and lack of job control were measured.

EMOTION WORK

373

It is only recently that authors tried to investigate the relationships between

more direct measures of emotional aspects at work and psychological strain (e.g.

Abraham, 1998; Adelmann, 1995; Brotheridge & Lee, 1998; Grandey, 1998;

Morris & Feldman, 1997). These authors referred to the concept of emotional

labour introduced by Arlie Russell Hochschild (1983). This concept refers to the

quality of interactions between employees and clients. The term “clients” is used

to refer to any person who interacts with an employee, for example, clients,

patients, children, customers, or guests. During face-to-face interactions with

clients many employees are required to express appropriate emotions as a job

requirement, for example, waiters or flight attendants are required to be friendly

even to arrogant or aggressive customers. Hochschild drew upon the work of

Goffman (1959) to argue that people in social interactions tend to play roles and

try to create certain impressions. Impressions include the display of normatively

appropriate emotions following certain display rules. In this respect, Morris and

Feldman (1996, p. 987) defined emotional labour as the “effort, planning, and

control needed to express organizationally desired emotions during interpersonal

transactions”.

It was our intention to investigate whether job requirements that refer to the

regulation of emotions could supplement the concepts of psychological job

stressors mentioned previously that refer to the regulation of cognition or

information processing. Hochschild, as a sociologist, differentiated between

emotional labour as the exchange value of work and emotional work as the use

value. In the present context, the psychological processes, for example, the

regulation processes of work actions rather than societal and economic aspects of

labour are considered. In psychology, the term “labour” is used to describe the

division of labour, labour–management relations, conflict resolution, and

collective bargaining. The term is not used when individual behaviour and

intrapsychic concepts are involved as in the concepts of physical and mental

work demands, work motivation, work involvement, work design, etc. To be

compatible with these research areas, the term “emotion work” is preferred. In

sum, emotion work possesses the following characteristics (Hochschild, 1983;

Morris & Feldman, 1997): (1) It is a significant component of jobs that require

either face-to-face or voice-to voice interactions with clients. This refers to the

service sector, in particular human services, but also to teachers, police,

correctional workers, debt collectors, and others. It should be noted that not all

jobs that require face-to-face interactions with clients belong to the service sector

and that defining service is problematic (Nerdinger, 1994). We will use the term

“person-related work” as an umbrella term for all jobs that require face-to-face or

voice-to-voice interactions with clients. (2) Emotions in these jobs are displayed

to influence other people’s attitudes and behaviours, usually by influencing their

emotional state. For example, a child nurse may show sympathy and talk to a hurt

child in a soft calming voice to make the child stop crying and cheer her up.

(3) The display of emotions has to follow certain rules. At present, many

374

ZAPF ET AL.

companies do not have explicit display rules as a part of the organizational

culture or as part of their job descriptions, in particular not in Continental Europe.

However, mission statements of companies sometimes incorporate display rules

and there may be implicit display rules taught in one’s occupational education or

as part of one’s professional ethos, for example, in the case of a nurse (Briner,

1995). In other cases, it may be the professional experience that you can’t sell

anything if you are not polite, friendly, and helpful. Employers differ in their

attempts to control and direct how employees display emotions to clients. In

some cases, it is part of the supervisors’ jobs to take care that display rules are

observed. Increasingly, companies ask customers to evaluate whether they were

treated in a friendly manner.

A number of studies operationalized emotion work as a dichotomous variable

indicating the presence or absence of emotion work in an occupation

(Hochschild, 1983; Wharton, 1993). Hochschild suggested that emotion work

depends on the frequency of interpersonal contact between employee and client,

thus conceiving emotion work as an unidimensional construct negatively

correlated with employees’ health. Accordingly, some authors (e.g. Adelmann,

1995) operationalized one scale for emotion work. However, these studies could

not find the expected negative relations between emotion work and psychological

strain, suggesting that more differentiated concepts should be used.

Other authors have worked on the differentiation of various aspects of

emotion work, many of them referring to the seminal work of Morris and

Feldman (1996). Some started with Hochschild’s (1983) concept of emotion

management to differentiate various dimensions of emotion work (e.g. Grandy,

1998; Kruml & Geddes, 1998). Other authors focused on determinants of

emotion work in the sense of “objective” job requirements: This emphasizes that

it is not in the discretion of the employee whether or not to express certain

emotions in a job. Rather, independent of a particular worker, it is required by the

organization and may be an explicit or implicit rule.

Approaches referring to the concept of emotion management differentiated it

based on how emotion work is done. One aspect differentiates between surface

acting and deep acting. Based on Goffman (1959), Hochschild (1983) argued that

individuals permanently manage their outer demeanour to conform with

situational requirements. Most emotion theorists propose that emotions consist

of several sub-systems (see Scherer, 1997): subjective feeling, physiological

reaction patterns, and expressive behaviour, the latter including facial

expression, voice and gesture. With reference to these concepts, surface acting

means that employees try to manage the visible aspects of emotions that appear

on the “surface” to bring them in line with the organizational display rules, while

the inner feelings remain unchanged. Another concept of Hochschild is “active

deep acting” when individuals try to influence what they feel in order to

“become” the role they are asked to display. In this case, not only the expressive

behaviour but also the inner feelings are regulated. Active deep acting refers to

EMOTION WORK

375

the case where an employee has to spend effort to regulate emotions. In other

cases, an employee may automatically feel the emotion required in a particular

situation. Hochschild called such forms “passive deep acting”.

Most studies of emotion work include the concept of emotional dissonance

(e.g. Abraham, 1998; Brotheridge & Lee, 1998; Grandey, 1998; Morris &

Feldman, 1996, 1997). Emotional dissonance occurs when an employee is

required to express emotions that are not genuinely felt in the particular situation.

A person may feel nothing when a certain emotion display is required, or the

display rule may require the suppression of undesired emotions and the

expression of neutrality or a positive emotion instead of a negative one.

Emotional dissonance may originate from “faking in good faith” when the

employee accepts the underlying display rule or from “faking in bad faith” when

the feeling rule is not accepted (Hochschild. 1983; Rafaeli & Sutton, 1987).

Various authors (e.g. Abraham, 1998; Adelmann, 1995) propose that faking in

bad faith has the most negative consequences.

Based on this, Grandey (1998) and Kruml and Geddes (1998) identified two

dimensions: emotional dissonance and emotional effort. Emotional dissonance

refers to Hochschild’s concept of surface acting and passive deep acting

(automatic emotion regulation), which are considered to be the opposite ends of a

continuum. If an employee spontaneously feels the emotion, emotional

dissonance is low; if he or she feels nothing or the opposite emotion, emotional

dissonance is high. Emotional effort refers to the degree to which employees

actively try to change their inner feelings to match the feelings they are expected

to express. According to Kruml and Geddes, this dimension incorporates

Hochschild’s (1983) active deep acting. Both dimensions showed a high

correlation in the studies of Grandey.

In conceptualizing emotion work as the behavioural response to variations in

the frequency, variety, intensity, and duration of interactions, Brotheridge and

Lee (1998, p. 7) used the term “emotional labour” to refer to “actions undertaken

as a means of addressing role demands”. In this sense, operationalizations of

emotion work come close to the concept of coping in stress research (Lazarus &

Folkman,1984; Semmer, 1996). The authors operationalized surface acting and

deep acting as the key constructs for emotion work. Deep acting refers to the

active attempts to align one’s felt and displayed emotion, which means that the

inner feelings have to be adapted to the emotions that have to be displayed. In

contrast, surface acting means pretending to have the emotions expected to be

displayed. In this case, employees do not try to feel the emotions they have to

display. Brotheridge and Lee considered surface acting as the manifestation and

even a proxy for emotional dissonance.

Morris and Feldman (1996) concentrated on what they called dimensions of

emotion work: the frequency of emotion display, the attentiveness to display

rules required (referring to the intensity and duration of emotion display), the

variety of emotions to be expressed, and emotional dissonance. They argued that

376

ZAPF ET AL.

all these dimensions of emotion work would increase emotional exhaustion, the

core variable of burnout.

To conclude: Most authors consider the frequency, variety, duration, and

attentiveness of emotions as dimensions of emotion work. Emotional dissonance

is viewed somewhat differently. Several authors consider it to be a result of the

determinants of emotion work (e.g. Adelmann, 1995); some authors even place it

close to the dependent variables (e.g. Ashforth & Humphrey, 1993). However,

there is some agreement to define emotional dissonance as the discrepancy

between displayed and felt emotions (Brotheridge & Lee, 1998; Morris &

Feldman, 1996, 1997) and to consider it as one of the key predictors of emotional

exhaustion. Brotheridge and Lee (1998) proposed that the emotional

requirements at work do not directly lead to emotional exhaustion but may do so

through their relation with emotional dissonance.

Several attempts have been made to operationalize aspects of emotion work.

Morris and Feldman (1997) operationalized three aspects of emotion work: the

frequency of emotion work, the duration of emotion work, and emotional

dissonance. For the frequency and duration scales they did not directly refer to

emotion display but referred to the frequency and time interacting with clients,

whereas emotional dissonance items directly referred to the match between

displayed behaviour and felt emotions. Brotheridge and Lee (1998) and Grandey

(1998) followed the model proposed by Morris and Feldman (1996) and

operationalized scales for the dimensions of emotion work as frequency, variety,

attentiveness, and duration (single item), and emotional dissonance, surface

acting, and deep acting as the core variables of emotion work. In the first study of

Brotheridge and Lee (1998), a factor analysis produced four factors collapsing

emotional dissonance and surface acting into one factor, and intensity, variety,

and duration into another. The two other factors were deep acting and frequency

of emotional display. In a second study, the authors were able to distinguish

frequency, variety, intensity, and duration, and surface acting and deep acting.

Best, Downey, and Jones (1997) measured how often different emotions were

expected on the job. Using factor analyses, they found three factors representing

the expression of positive emotions, suppressing negative emotions and

expressing negative emotions, whereby the latter showed a low reliability and a

low response frequency. Abraham (1998) operationalized emotional dissonance

using items from Adelmann (1995) that referred to display rules in the

organization. She then developed identical items rephrased to reflect the degree

to which the respondents would actually show the corresponding emotions.

Difference scores of the respective items were then computed to reflect

emotional dissonance.

For the present studies we combined the literature on emotion work described

previously with action theory-based approaches in stress research (Frese & Zapf,

1994; Greiner & Leitner, 1989; Zapf, 1993). “Work” or “labour” is a

multidisciplinary concept. Hacker (1973, 1998) and Volpert (1974) argued that

EMOTION WORK

377

the psychological component of work is the work activity and from the

perspective of action theory it is the psychic regulation of work actions. Through

various cognitive processes, action theory links the objective work environment

to behaviour. To describe job requirements, three aspects are distinguished: the

regulation requirements of a task, regulation possibilities, and regulation

problems (for details, see Frese & Zapf, 1994; Zapf, 1993).

From an action-oriented perspective, regulation requirements are related to

properties of the hierarchic-sequential organization of action and comprise the

complexity of decisions, the number and connectedness of goals and sub-goals,

and the extent of conscious vs. automatic regulation processes. Regulation

possibilities refer to the concept of control. Control means having an impact on

one’s conditions and on one’s activities in correspondence with some goal.

Decision possibilities exist with regard to the sequence of the action steps, the

timeframe, and the content of goals and plans (Frese, 1987). Several authors have

operationalized various aspects of control, such as task control referring to

decision possibilities regarding the goals to be carried out, the sequence of plans

to be performed, and the sequence of feedback information processing. Time

control, for example, refers to both when and for how long a certain task is

performed (e.g. Frese & Zapf, 1994, Semmer, et al. 1995; Wall, Jackson,

Mullarkey, & Parker, 1996; Zapf, 1993). Regulation problems are an action

theory conceptualization of work stressors. The stressors are differentiated

according to how they disturb the regulation of actions (Frese & Zapf, 1994;

Greiner & Leitner, 1989; Semmer, 1984).

There is evidence that regulation requirements, regulation possibilities, and

regulation problems are differentially related to health and well-being and that

this differentiation helps to overcome a stimulus-response framework where

every characteristic of the job has negative consequences and where “doing

nothing” would be the best concept to avoid stress at work. In contrast, action

theory proposes that human beings usually try to actively cope with their

environment. In this sense, job design should support this active approach by

providing challenging (i.e. sufficiently complex) tasks (regulation requirements)

and control (regulation possibilities), but at the same time, reducing the stressors

(regulation problems). Regulation requirements are relevant to the concept of

personality enhancement (Hacker, 1973, 1998; see also Frese & Zapf, 1994).

This means that they enable one to develop cognitive and social skills, and further

satisfaction and self-esteem. They follow the person–environment fit model

(Edwards & van Harrison, 1993): They are positive as long as they are matched

by personal prerequisites and they become negative when they exceed them.

Research shows that regulation possibilities (control) typically show a direct

positive effect as well as a moderating effect between stressors and strains (e.g.

Kahn & Byosiere, 1992). In contrast, regulation problems (stressors) have

negative health effects. Stressors are in a sense independent of the person–

environment fit, because people want challenging tasks, but they do not need a

378

ZAPF ET AL.

minimal amount of conflicts, time pressure or superfluous organizational

problems to feel happy.

Using an action theory framework, the psychological focus of the present

study was on the regulation of emotion display according to a goal given by the

organization. In this sense, emotion work is part of intentional and goal-directed

behaviour. From the organization, an employee receives an order to carry out a

certain task in a certain way. This includes behaving according to the emotional

display rules of the organization. The order is then redefined into a subjective

goal (Hackman, 1970). Emotion work usually refers to a sub-goal of a higher

order goal and requires certain emotion display during an interaction with a

client. Ideally, emotion work is done in the automatic mode, that is, the emotion

is automatically shown in the social interaction as required (cf. Scherer &

Wallbott, 1990). In this sense, the concept of emotion work differs from

approaches that investigate emotions as a response to a variety of organizational

conditions (e.g. Basch & Fisher, 1998).

Emotion work poses various demands on the worker. This view considers the

job requirement aspects of emotion work and is congruent with the idea that

objective job characteristics or job stressors created by the organization affect the

workers in various ways (Frese & Zapf, 1988, 1994; Spector, 1992). This

approach conforms to the behaviour requirement approach in job analysis

research (Hackman, 1970). Because our goal was to develop an instrument that

should be used in addition to other instruments in the analysis of stress at work,

we did not intend to operationalize all aspects of emotion work separately, but we

used these concepts for the development of items. However, because of the

empirical findings of Best et al. (1997), Brotheridge and Lee (1998), and

Grandey (1998) we expected that the items of emotional requirements would

represent at least two factors, namely the frequency and the variety/intensity of

emotion work. We did not model the duration aspect. The reason is that the

emotion work components were developed in the context of stress research

where the frequency of emotion work seems to be most relevant. If intensity only

is measured there should not necessarily be a strong relation with variables such

as burnout, because the more intense emotions could be more seldom. Similarly,

if variety of emotions is measured, the problem occurs that a high variety in

general might be more stressful than a low variety but it may not occur very often.

In the present study we partly tried to circumvent this problem by asking, for

example, how often both positive and negative emotions have to be displayed.

In addition to the work of Hochschild (1983), we drew upon concepts of

emotion work that put the influence and management of clients’ emotions into

the foreground (e.g. Brucks, 1998; Strauss, Farahaugh, Suczek, & Wiener, 1980;

Strazdins, 1998). To be able to manage clients emotions, the accurate perception

of the clients’ emotions is an important prerequisite. This is also in accord with

communication psychology (Riggio, 1986) and the literature on emotional

intelligence (Goleman, 1995, 1998). Riggio operationalized basic social skills

EMOTION WORK

379

that are related to the regulation of emotions and differentiated sensitivity,

expression, and control of emotions. Expression and control refer to the

emotional requirements described previously. In addition, we operationalized

“sensitivity requirements” as the necessity to be sensitive and consider the

emotions of clients. It can be expected that sensitivity requirements are positively

correlated with emotional requirements because the expression of an emotion

during an interaction usually is dependent on the emotion of the interaction

partner. Only in short script-like interactions might a person express emotions

without trying to sense the emotion of others.

In his qualitative study on supermarket clerks’ performance, Tolich (1993)

argued that the presence or absence of control over one’s emotion display is one

of the important issues of emotion work. He differentiated regulated emotion

management from autonomous emotion management. Referring to the dif-

ferentiation of various aspects of job control (regulation possibilities) mentioned

earlier, emotion work control was operationalized as a special case of job control

with regard to the display of emotions (in the sense of Tolich’s autonomous

emotion management) and interaction control as a special case with regard to the

underlying social interactions where emotions have to be displayed. Emotion

work control refers to the extent to which an employee can decide whether or not

to show a desired emotion. Emotion work control is probably lower when display

rules have been made explicit in an organization, but this should not be

necessarily so. Waitresses in restaurants may have to follow certain display rules,

but there may be differences in how often and in what cases the waitresses are

empowered to deviate from the rules. As described previously, some authors

have operationalized emotion work by operationalizing aspects of the underlying

social situation (e.g Adelmann, 1995; Morris & Feldman, 1997). In a similar way

we operationalized the control of the social interaction, that is the degree of

influence an employee has in social interactions with clients. An example is

whether an employee can decide when to stop an interaction with a client. There

are several reasons why we included the concepts of emotion work control and

interaction control. First, they are part of the action theory framework we applied

to emotion work. Second, qualitative research done by Hochschild (1983),

Rafaeli (1989) and Tolich (1993) pointed to the importance of this concept.

Third, the study of Erickson (1991, cited in Abraham, 1998) showed some

evidence that the moderating effect of job control applied when emotional

dissonance is involved and that this effect might be even stronger when the

control concept is matched to the stressor (cf. the analogy of the match-

hypothesis of stressors and social support of Cohen & Wills, 1985).

Finally, emotional dissonance was considered as an emotion regulation

problem. As in most of the other approaches, it is defined as the mismatch

between felt emotions and the organizationally desired expression of these

emotions. We considered emotional dissonance as an external demand rather

than a reaction to emotion display or a behavioural strategy. One could argue

380

ZAPF ET AL.

that, given a certain requirement for frequency and content (positive or negative

emotion), it should then depend on the employee and his or her personality to

what extent he or she feels in line with the required emotions. In this sense,

emotional dissonance would be a stress reaction and a first sign of emotional

exhaustion. However, there are qualitative differences in social situations that are

not sufficiently described by the parameters for display rules. This is because the

display rules describe the desired state of emotion display, but they do not

comprise anything about how often individuals are exposed to situations where

they have to show the required emotions. Moreover, they do not reflect other

factors, namely how positive or negative the social interaction is, which may

influence what people feel and whether this fits to the emotion required by the

display rule for this particular situation. Compare, for example, a nurse in a

children’s hospital and a nurse in a retirement home. The display rules of

showing friendliness and empathy may be the same, and frequency and duration

of interactions may be similar, leading to similar required display rates of

positive emotions, but the nurse in the retirement home may encounter many

more situations where an average person feels disgust or anger. Similarly,

cashiers of a supermarket chain may all have the same requirements to display

positive emotions to customers, and the number of customers determining the

frequency of emotional requirements may be similar. However, depending on

where a supermarket is located, there may be differing frequencies of encounters

with complaining or otherwise negatively behaving customers, which is a good

predictor of negative emotions of the employee (Doucet, 1998). Consequently,

the number of situations where gaps between felt and desired emotions appear

may differ considerably. The discrepancy between what an average person is

likely to feel and what the respective display rule is, varies from situation to

situation. Therefore, the aspects covered by the concept of emotional dissonance

are not covered by the frequency and other parameters of emotional requirements

because they all refer to the display rules and to more formal characteristics of

social interaction, such as frequency and duration, and not to the quality of the

actual situations and the resulting differing discrepancies between display rules

and average emotions in a given situation.

Two more issues should be mentioned with regard to emotional dissonance.

First, some authors focus on the display of emotions required by the organization,

no matter what a person feels (e.g. Ashforth & Humphrey, 1993). Display of

emotion refers to facial expression, bodily behaviour, and voice. These are the

visible aspects of the emotional system (Scherer, 1997; Scherer & Wallbott,

1990). If this is so, could then the display of emotions not be described by

sensorimotor processes? The regulation of sensorimotor processes is, for

example, a part of the action theory approach mentioned previously (Frese &

Zapf, 1994; Hacker, 1998). According to action theory, sensorimotor processes

are highly automatized. They are usually carried out in the automatic mode, that

EMOTION WORK

381

is, without conscious attention. This is also so with respect to the sensorimotor

processes in the expression of emotion (Ekman, 1984; Izard, 1977; Scherer &

Wallbott, 1990). If a certain emotion is felt, then the expression of this emotion

automatically occurs whereby social competence may play a moderator role. If

an emotion that has to be displayed, is not felt, then problems occur. In highly

standardized situations it may be easy to fake. If this is not the case, then the true

feelings may show through and may be recognized by other people (cf. Ekman &

Friesen, 1982 who investigated the differences between true and faked smiling).

In some cases, authenticity, that is not faking, may even be a key variable, for

example, for therapists in encounter therapy (Rogers, 1951). Hochschild (1983)

also raised the problems of surface acting and discussed that even in services

such as airlines deep acting is required. Employees are required to feel the

emotions they should feel because otherwise there is the danger that it would not

work. Findings in social psychology showed that people can sometimes tell when

someone is faking a friendly face (Ekman & Friesen, 1982). Therefore,

emotional dissonance as a job stressor should lead to both surface acting and deep

acting as a reaction, implying that emotion work cannot be reduced to the

sensorimotor regulation of emotional expression. This also shows the difference

between emotional dissonance as a stressor and as a reaction. If deep acting was

successful there is no internal state of emotional dissonance (emotional

dissonance as a reaction or dimension of emotion work). However, deep acting

can be a strategy to deal with the job stressor of emotional dissonance.

Finally, emotional dissonance has to be discussed with reference to the

differentiation of use value and exchange value of work (Hochschild, 1983;

Marx, 1867/1977; Nerdinger, 1994). Hochschild (1983), who coined the term

“emotional labour”, pointed out that “emotional labour is sold for a wage and

therefore has exchange value” (p. 7). Nerdinger discussed in detail that, from the

economic point of view, the work of the service provider is exchanged for

money. However, in many cases the full service requires an interaction as if there

were not an economic but a family-like relation. A therapist is expected to be

truly interested in the client and not just because he or she is paid for it. Similarly,

parents know that child nursing is a job and that the nurses work for money.

Nevertheless they wish that the nurses really love their children. Nerdinger

pointed out that the social interaction is not only a means to deliver the service but

is part of the service product. Thus, a service employee may face contradictory

expectations given by the personal interaction with the client (who, for example,

may want advice) and the economic interests of his or her employer (who may

insist on high sales). Moreover, the requirements of the organization itself may

be ambiguous. A computer hot-liner may be required to be customer friendly,

but, at the same time limit talks with customers to 5 minutes. One can

hypothesize that such contradictory job requirements are a source of emotional

dissonance in any kind of person-related work.

382

ZAPF ET AL.

In sum, applying the concept of action theory to emotion work first leads to

integrating the special control concepts described earlier. Second, it helps to

understand that emotion work is not necessarily negative but has also positive

implications. To explore the construct validity of the instruments developed for

the present studies, we developed several hypotheses.

Hypothesis 1: Emotion work is a multidimensional

construct

Within emotion work, emotional requirements, emotion control, and emotional

dissonance can be distinguished. The differences in these concepts have been

described previously. Empirically, this hypothesis is supported by most findings

in the literature so far with regard to emotional requirements and emotional

dissonance (e.g. Brotheridge & Lee, 1998; Grandey, 1998; Morris & Feldman,

1997). We are not aware of studies that operationalized a concept equivalent to

emotion work control or interaction control. Also, the fact that studies that did

not differentiate between various aspects of emotion work did not find the

expected results (Adelmann, 1995), supports the view that emotion work is not a

homogeneous construct.

Hypothesis 2(a): Emotional requirement scales are positively

correlated

It is expected that the various aspects of emotional requirements are highly

correlated, whereas emotional requirements and emotional dissonance should

show a positive but lower correlation. The reason is that all emotional

requirements are a function of the interaction time with clients and the existence

of display rules. In contrast to Morris and Feldman (1996), most of the items of

the present study have a frequency component. The more interactions a person

has with a client, the more this person is supposed to show positive and negative

emotions. In addition, the sensitivity requirements should also be high.

Sensitivity requirements should be positively correlated with the other scales

because the expression of emotion should in most cases be dependent on the

emotions of the interaction partner,which have to be adequately perceived.

Hypothesis 2(b): Emotional requirement scales are also

positively correlated with emotional dissonance

The frequency of emotional dissonance also depends on the frequency of

interactions. Therefore, emotional dissonance should be positively correlated

with the emotional requirement scales. However, because emotional dissonance

is also a function of how pleasant or unpleasant the social interactions are, the

correlations are expected to be lower than the correlations amongst the emotional

requirement scales, which all mainly depend on the interaction frequency. In

EMOTION WORK

383

addition, Morris and Feldman (1996) proposed that the higher the frequency of

emotion display, the higher is the chance that emotions have to be displayed that

do not fit the emotions felt. A similar argument applies for the variety of

emotions. For variety of emotions it can be added that it is more likely that

employees have problems with negative emotions compared to positive

emotions. With regard to the correlation between emotional requirement

variables and emotional dissonance, the empirical findings are mixed. Grandey

(1998) found a correlation between suppressing negative emotions and

emotional dissonance, but not between expressing positive emotions and

emotional dissonance. Also, Brotheridge and Lee (1998) found a correlation

between frequency of emotion display and emotional dissonance.

Hypothesis 2(c): Emotion control is negatively related with

emotional dissonance

There is evidence in the stress literature that control is negatively related to job

stressors. Abraham (1998) was able to demonstrate such a relationship for job

autonomy and emotional dissonance. This negative relation should also occur for

specific control measures such as emotion work control and interaction control,

especially if the specific control variables match the specific kinds of stressors,

which should be the case in the present study.

Hypothesis 3: There are both positive and negative

relations between emotional requirements and strain

and well-being

Much of the literature addressed the negative effects of emotion work (e.g

Adelmann, 1995; Hochschild, 1983). Most often scholars cited the negative

relations with burnout, hypothesizing that emotion work would increase

emotional exhaustion and depersonalization and would reduce personal

accomplishment. Some authors discussed relationships with poor self-esteem

and depression. Hochschild, in particular, referred to the problem of alienation

from one’s true feelings. A few authors, however, also referred to potential

positive effects such as job satisfaction, self-esteem, and self-efficacy (e.g.

Ashforth & Humphrey, 1993; Morris & Feldman, 1997; Stenross & Kleinman,

1989; Tolich, 1993; Wharton, 1993).

Drawing parallels with action theory-based concepts of work characteristics

(Frese & Zapf, 1994; Hacker, 1998), it can be assumed that work in general is not

either positive or negative. Rather, a challenging job may comprise positive job

content variables such as job control, complexity, or variety at work, but

challenging jobs often go along with high quantitative workload and uncertainty

at work. Similarly, it is not assumed that emotion work is generally either

negative or positive. On the one hand, emotion work is laborious and effortful.

Therefore, if a high frequency of emotional display and a variety of emotions are

384

ZAPF ET AL.

required this should lead to psychological strain, especially to emotional

exhaustion (Morris & Feldman, 1996). When emotional requirements exceed

certain limits the likelihood increases that the emotions that have to be expressed

do not match the emotions that are felt at that moment. That is, in line with the

person–environment fit model (e.g. Edwards & van Harrison, 1993), if emotional

requirements are frequent and last for a long time, their effects on well-being

should be negative. This assumption is supported by findings in the burnout

literature. Maslach (1982) stated that frequent, intense, and charging face-to-face

interactions were associated with higher levels of emotional exhaustion. Cordes

and Dougherty (1993) in their review reported that longer interactions with

clients were associated with higher levels of burnout. Morris and Feldman (1997)

considered emotional exhaustion as the key consequence of emotion work.

Reviewing the empirical literature, however, shows that the expected correla-

tions between emotional requirements and emotional exhaustion were often not

(Adelmann, 1995; Morris & Feldman, 1997) or only occasionally found

(Brotheridge & Lee, 1998; Grandey, 1998). For Morris and Feldman, this may be

due to the fact that they did not directly refer to the frequency or duration of

emotional display but to the underlying social interaction. Although empirical

findings are mixed here, it is assumed that variables representing emotional

requirements are positively correlated with emotional exhaustion and other

variables of psychological strain.

Based on the literature on the affiliation motive it can be assumed that dealing

with other people and expressing emotions when interacting with these people

satisfies affiliation, status, and recognition needs, for example, by showing

altruistic behaviour (e.g. Bierhoff, 1990; Hill, 1987). In many cases, the

expression of emotion can be thought of as a spontaneous process experienced

not to be effortful at all (cf. Ashforth & Humphrey, 1993; Scherer & Wallbott,

1990), but contributing to a social situation with positive consequences for the

employee concerned. The intentional expression of positive emotions usually

increases the probability of the interaction partner to show reciprocal positive

emotions in return (Wiemann & Giles, 1997). This can be perceived as positive

feedback contributing to the employee’s satisfaction and self-esteem. Adelmann

(1995) referred to the facial feedback hypothesis to argue for positive effects of

emotion work. There is at least some evidence for the weak form of this

hypothesis: In an experiment, Strack, Stepper, and Martin (1988) showed that

participants whose muscle groups necessary for laughing were stimulated found

a movie more funny in comparison to a group whose laughing muscles were

inhibited.

There is, indeed, some evidence of the positive implications of emotion work.

On a qualitative level, Tolich (1993) described supermarket clerks who enjoyed

showing prescribed emotions in the form of jokes or entertainment of customers

who chose their checkout lines. Stenross and Kleinman (1989) reported that

detectives positively assessed interrogations with criminal suspects because this

EMOTION WORK

385

played a central role for goal achievement, namely, solving a case. Wharton

(1993) found a positive relation with job satisfaction and Grandey (1998)

reported a positive correlation between expressing positive emotions and job

satisfaction. Based on these considerations and empirical findings it was

expected that emotional requirement variables are positively correlated with

personal accomplishment, self-esteem, and job satisfaction.

Hypothesis 4: Emotion work control and

interaction control have a positive effect on health

Emotion work-related control is conceptualized as a special case of job

control concerning the possibility to decide whether or not one likes to express

emotions in a certain situation. In many studies on job stress (e.g. Kahn &

Byiosiere, 1992) it has been shown that job control is typically positively related

to well-being. Therefore, it can be expected that this is also true for the special

cases of emotion work control and interaction control. There is no direct

empirical evidence so far. However, on a qualitative level, it has been shown that

the exertion of control in social interactions was perceived to be positive and that

employees struggled for control in interactions with clients (Rafaeli, 1989;

Tolich, 1993).

Hypothesis 5: Emotional dissonance is negatively

related with health

Hochschild (1983) was the first who described the negative effects when positive

emotions have to be displayed when either nothing is felt or if the felt emotions

are even in contrast to the displayed emotions. Hochschild asserted that if

employees do not feel what they ought to feel, they may blame themselves and

feel phony and hypocritical. This may result in low self-esteem (Kruml &

Geddes, 1998). In such cases, they may also start to blame the company, which is

likely to go along with decreased job satisfaction. Rafaeli and Sutton (1987)

argued that emotional dissonance is a form of person–role conflict (Kahn et al.,

1964), which means that one has to do things that are against one’s better

judgement. Because role conflict is a strong predictor of emotional exhaustion

(cf. the meta-analysis of Lee & Ashforth, 1996), it was hypothesized that

emotional dissonance is also a strong predictor of exhaustion. All in all, the

clearest empirical relation between emotion work variables and psychological

strain occurred for emotional dissonance and emotional exhaustion (Abraham,

1998; Brotheridge & Lee, 1998; Grandey, 1998; Morris & Feldman, 1997) and

depersonalization (Grandey, 1998). Mostly, no relationship was found for

relations with personal accomplishment (Grandey, 1998). Empirical evidence on

the relation between emotional dissonance and job satisfaction is mixed. Morris

and Feldman found a negative relation between emotional dissonance and job

satisfaction, whereas Grandey (1998) did not.

386

ZAPF ET AL.

METHOD

Samples

The following analyses were based on three samples. The first sample consisted

of employees working in a home for handicapped children and other social

service institutions (N=83) in South Germany: 80% were women, which is

typical for the human services; average age was 38 years; 58% worked in the

children’s home and had direct client contact (nurses, team leaders, social

workers), 19% worked also in the children’s home, but had no direct client

contact (administration, technical staff). Moreover, 24% were employees with

direct client contact in a hospital, and in other homes for problem children and

handicapped children. As in the other studies, participation was voluntary.

The second sample was collected in the hotel business. With the help of the

Berufsgenossenschaft Nahrung (professional food association), 27 hotels were

contacted and 867 questionnaires were sent off. We received 175 questionnaires,

which corresponded to a response rate of 20.3%. Discussion with our personal

contacts with hotel representatives revealed that the low response rate was

mostly due to heavy workload of the hotel employees. Most of the participants

were employed in hotels as front-line officers, waiters, or waitresses and other

hotel professions; 71% were women and 80% of the sample were between 18

and 32 years old (25% between 18 and 22, 32% between 23 and 27, and 24%

between 28 and 32 years); 15% had a general secondary school degree

(Hauptschule), 29% had a lower school-leaving certificate (mittlere Reife), 44%

had a high school diploma (Abitur), and 11% some type of university degree.

Only a minority of 18% did not have a special occupational education related to

the hotel business.

The third sample consisted of 250 participants employed in 14 call centres of

various firms with an overall response rate of 50%. This sample consisted of

employees who all had voice-to-voice contacts with clients: 75% were female;

the average age was 31 years; 74% had a high school diploma (Abitur) or some

kind of university degree; 76% received calls but did not call clients themselves

(inbound), the others mostly both called clients and received calls (inbound and

outbound; details in Isic, Dormann, & Zapf, in press).

Procedure

In Study 1, we first developed a list of items covering the constructs of emotion

work described earlier. Then we gave a first draft of the questionnaire to nurses

and social education workers who gave feedback with regard to the applicability

and comprehensibility of the items. This process was repeated and the resulting

questionnaire was administered to the first sample. Starting with this version and

the first empirical results of Sample 1, a version of the questionnaire was

developed for the hotel business sample. During this process it became clear that

EMOTION WORK

387

part of the developed items were domain specific, whereas another part seemed

to be applicable for a variety of professions. In the hotel business, the first draft of

the questionnaire was discussed with 10 persons from various service branches.

They received the questionnaire and were also interviewed. A revised version

was then applied to a sample of hotel service students (n=26). These students

were asked to fill in the questionnaire and to comment on the applicability and

comprehensibility of the items. These comments and the psychometric results led

to the final version used for the hotel sample. This version consisted of domain

specific and general items. The general items of Sample 2 and some newly

developed general items were applied in the third sample. Here, the questionnaire

was much shorter than in the previous versions, because it was intended to

develop a short instrument which can be applied in all areas with person-related

work and which can be used as a supplement to other job stress instruments.

Instruments

To assess the construct validity of emotion work, several other variables were

included in the study.

Job satisfaction was measured by the Kunin-item in Study 2. According to

Wanous, Reichers, and Hudy (1997), a reliability of 0.57 was assumed for this

single-item measure. In the other studies a job satisfaction scale developed by

Semmer and Baillod, (1991) was used. It consisted of the Kunin item and various

other items which were developed on the background of the job satisfaction

model of Bruggemann (1974). High scores mean satisfaction, low scores mean

dissatisfaction including a resigned attitude towards one’s job.

Psychosomatic complaints, irritation, and self-esteem were measured using

scales developed by Mohr (1986, 1991). The psychosomatic complaints scale

consisted of a list of 20 psychosomatic symptoms such as nervousness,

headaches, tension, high blood pressure, and insomnia. Irritation consisted of

items referring to anger and not being able to stop thinking about one’s work.

Finally, self-esteem was measured with items like “I am proud of my

achievements”. The pychosomatic complaints and self-esteem items were

answered on a 5-point scale, whereas a 7-point scale was used for irritation.

Burnout (emotional exhaustion, depersonalization, and personal accom-

plishment) was measured using the Maslach Burnout Inventory—German

version of Büssing and Perrar (1992). Emotional exhaustion measures one’s

feeling of being burnt out, frustrated, and perceiving working with people to be

very demanding. Depersonalization comprises the tendency to treat clients like

objects and to become indifferent and apathetic with regard to clients. Reduced

Personal Accomplishment includes the feeling of having reduced energy to do

388

ZAPF ET AL.

things and of not being able to meet one’s aspirations. The burnout items were

answered on a 7-point scale. The descriptive data of the scales used for validation

purposes are summarized in Table 1.

RESULTS

In the first study, because of sample size, we used exploratory factor analyses

EFA (principal components analyses and varimax rotation) to test whether the

items behaved as predicted by the theoretical concept. In the second and third

study we applied confirmatory factor analyses CFA using LISREL 8.3 of

Jöreskog and Sörbom (1993), however, also in an exploratory manner. The

results of Study 1 were used as starting models.

For Study 1, EFA showed that the items for emotional requirements loaded on

three factors. The first factor referred to the display of positive emotions

(example item: “How often does it occur in your job that you have to express

pleasant emotions towards your clients?)”. The second factor comprised items

referring to the variety of emotion display and the need to deal with negative

emotions of clients (example items: “How often does it occur in your job that you

have to express unpleasant emotions towards your clients?” “How often does it

occur in your job that you have to consider negative moods of your clients?”). As

expected, “sensitivity requirements” also led to a separate factor (example item:

(“Does your job require you to pay attention to the feelings of your clients?”).

This factor consisted of items that asked whether sensitivity or knowledge about

the clients’ current feelings is a job requirement. Moreover, a factor for

emotional dissonance appeared consisting of items referring to displaying

emotions not felt as well as to the suppression of felt emotions (example item:

“ ‘A’ can openly display his/her feelings towards clients—‘B’ has to display

feelings towards clients which do not match his/her true feelings. What is your

job like?”). Finally, a factor comprising items referring to control with regard to

social situations where emotion work is taking place was developed (example

item: “Is it up to you how long you pay attention to a client?”). Contrary to our

intention it was not possible to develop a scale for emotion work control. The

items of this scale loaded on other factors as well, particularly on the emotional

dissonance and the interaction control factor.

In Study 2, we used CFA for scale development. We started with the solution

of Study 1 and tried to model a positive emotion display factor and a negative

emotion/variety factor. Moreover, we again tried to model a factor for emotion

work control. The first attempts showed a low fit. Again, it was not possible to

develop a factor for emotion work control, but for interaction control. Second, it

turned out that the emotional requirement items fell into four groups: positive

emotions display, negative/variety of emotions, sensitivity requirements, and

items which referred to showing sympathy as a job requirement. The inspection

of the sympathy items showed that these items were difficult to locate on a

EMOTION WORK

389

rit

lf-

tio

iza

tio

ish

tis

l e

sti

liz

ish

–0

Irr

ita

tio

–0

lf-

–

–0

–

–0

tis

–0

’s

; *

t o

f 3

n’

s h

e (

; r

: h

s (

);

a s

e K

-it

; a

s e

t a

, e

sti

lia

lit

390

ZAPF ET AL.

(

)

sit

tiv

iti

ire

t t

s p

t e

–0

rin

sto

e f

lin

rfi

t t

s u

t e

iff

t e

sit

tio

s o

t o

r u

t e

t t

–0

rta

isp

t c

isp

sit

e e

ffe

–0

f t

rta

t t

iti

e f

lin

f c

–0

rta

t a

sto

rta

e c

’s

iti

fte

r a

t t

e p

sib

e e

s w

ith

sto

s a

s s

ila

rru

ith

sto

ith

sto

t o

sto

s’

ish

s o

e c

’s

e i

tio

t o

f t

sto

’s

f =

) =

, R

t m

r o

f a

, G

s o

f f

it)

, N

t i

) =

-n

t i

) =

390

EMOTION WORK

391

, s

, a

ria

sit

tiv

iti

ire

iss

isp

sit

–

iti

ire

–

l d

iss

–

–0

–

–0

–

–0

–

–0

–

–0

’s

; *

t o

f 3

n’

s h

; r

: h

s (

);

392

ZAPF ET AL.

positive-negative emotions dimension. Showing positive emotions obviously

meant to show emotions to make clients feel happy. Showing sympathy meant to

feel with a client who for some reason feels negative. Showing negative emotions

referred to negative social interactions, for example, talking to a guest who was

molesting another guest. Little problems occurred for the modelling of the

emotional dissonance and sensitivity requirements factors. Finally, it was

possible to model the interaction control factor. We stopped model trimming

when the factor loadings were significant and when there were theoretical

reasons for not eliminating items with cross-loadings.

Finally, in the third sample, a positive emotions display and negative

emotions/variety factor, sensitivity requirements, emotional dissonance and

interaction control could be modelled. In addition, there was a factor “routineness

of emotional display" consisting of two items. The emotion work control factor

could, again, not be modelled. The result of the confirmatory factor analysis of

Study 3 is shown in Table 2. Using the CFA procedure in an exploratory manner

we used the modification index of LISREL for model trimming. The

modification index indicates where model restrictions do not fit with the data. It

is suggested that modification indices higher than 5 indicate a substantial misfit

(Jöreskog & Sörbom, 1993). Therefore, we included all cross-loadings indicated

by modification indices higher than 5. This procedure led to six cross-loadings of

which two are relevant. The intensity of emotions item also showed a high

negative loading on the routineness factor indicating that intensity might also be

considered different from the frequency of positive emotion display (cf. the

results of Brotheridge & Lee, 1998). Second, expression of different emotions

also showed a high loading on the positive emotion display factor. Table 3 shows

that it was possible to develop the scales for emotion work with satisfactory

reliabilities (Hypothesis 1). Although some cross-loadings occurred in all three

studies we consider discriminant validity to be satisfactory.

According to Hypothesis 2a we expected that the emotional requirement

scales would show significant positive correlations. This was so in all cases with

correlations ranging from 0.22 to 0.63. Second, it was also expected that the

emotional requirement scales would also be positively correlated with emotional

dissonance (Hypothesis 2b). This was the case for Studies 2 and 3; for Study 1

the correlations were in the expected direction with two of the three correlations

significant at the 10% level. Finally, according to Hypothesis 2c, emotional

dissonance was expected to be negatively correlated with interaction control.

This was the case in Study 3. In Study 1 the correlation failed to reach

significance level, while in Study 2 the correlation was not significant.

The third hypothesis was that the emotional requirement scales would show

positive and negative correlations with strain and well-being. With one exception

(display of positive emotions in Study 2), all emotional requirement scales were

correlated with emotional exhaustion (Table 4). In Study 3, all emotional

requirement scales were also correlated with depersonalization. For Study 2, this

EMOTION WORK

393

-b

sit

tiv

iti

ire

l e

sti

–

–.

–

–.

–

–.

liz

–

–.

ish

–

–.

–

Irr

ita

tio

–

–.

–

–.

–

–.

–

–.

lf-

–.

–

–.

–

–.

–

–.

tis

–

–.

–

–.

–

–.

–

–.

. F

t o

f 3

ild

n’

s h

e (

; r

: h

s (

);

393

394

ZAPF ET AL.

was the case in three of four cases. For Studies 2 and 3, sensitivity requirements

were positively correlated with irritation and psychosomatic complaints. In

addition, display of positive emotions and sensitivity requirements were

positively correlated with personal accomplishment, indicating the positive

effects of emotion work. However, most of the correlations with self-esteem and

job satisfaction were not significant.

In Hypothesis 4 we assumed negative relations between control and

psychological strain. This hypothesis was mostly rejected because it was not

possible to develop a scale for emotion work control. For interaction control, 6

out of 21 correlations were significant and in the expected direction.

Finally, as expected, emotional dissonance showed the clearest effects on

psychological strain. In all three samples, high correlations appeared with

emotional exhaustion, depersonalization, irritation, and psychosomatic

complaints. In addition, in two of the three samples, the expected negative

correlation with job satisfaction appeared.

DISCUSSION

In this article we argued that traditional job stress analysis instruments are

not able to cover the full range of job stressors frequent in work where interaction

with clients is a central part. We described the concept of emotion work,

first introduced by Hochschild (1983), as the job requirement to display

organizationally desired emotions. Following action theory-based conceptions

of job analysis, we differentiated between emotional regulation requirements,

emotional regulation possibilities, and emotional regulation problems. Based

on three empirical studies in a handicapped children’s home, in the hotel

business, and in call centres, we were able to develop scales for the requirement

to display positive emotions, the requirement to display negative emotions

including also a high variety of emotions, and sensitivity requirements. In

one study, a factor “requirement to show sympathy” was found. In all three

studies we were able to develop scales for interaction control, but not for emotion

work control. Finally, a scale for emotional dissonance could be developed. The

scales showed satisfactory reliabilities. Discriminant validity could be

demonstrated, although some items showed cross-loadings on other factors. This

should be improved in future studies. Most of the hypotheses regarding construct

validity were supported by the data. Emotional regulation requirement scales

were both positively and negatively related to various variables of psychological

strain and well-being, supporting the view that person-related work is—as is

object-related work—not negative per se. As in most of the other empirical

studies, emotional dissonance has proven to be a stressor that shows negative

relationships with health. Finally, interaction control partly showed the expected

negative correlations with psychological strain and positive relations with job

satisfaction.

EMOTION WORK

395

Several issues are noticeable in the present studies. First, the studies support

the proposition that emotion work is a multidimensional phenomenon (e.g.

Morris & Feldman, 1996). They show that sub-constructs of emotion work have

partly contradictory relations with health and well-being, thus explaining the

failure to find correlations with dependent variables when overall scales for

emotion work were used (e.g. Adelmann, 1995).

Expressing positive emotions showed both positive and negative relations

with health variables. The positive correlations of the requirement to express

positive emotions with personal accomplishment supports the view that

Hochschild’s proposition of emotion work to be alienating and stressful is one-

sided (cf. Stenross & Kleinman, 1989; Tolich, 1993; Wharton, 1993). It is likely

that the expression of positive emotions is reinforced by positive customer

reactions and is generally valued both by the customers and the management

(Doucet, 1998). On the other hand, the failure to find consistent correlations with

job satisfaction and self-esteem obviously requires further research. It could

either be because of methodological problems or because the emotional

components of social interactions are of significantly less importance compared

to cognitive task aspects. Variety/negative emotions showed more negative

relations compared to the display of positive emotions. This supports the

qualitative findings of Stenross and Kleinman (1989), who found that showing

sympathy and dealing with negative emotions of others was a stressful

experience for detectives. On the other hand, it can be assumed that dealing with

complicated interactions where negative emotions have to be handled can also be

a source of feelings of personal accomplishment when this is considered to be a

genuine part of one’s job, as in the case of employees of the handicapped

children’s home.

In all, it is remarkable that all our studies differentiated between the display of

positive emotions and a factor either referring to the variety of emotions or to

displaying or treating negative emotions. Obviously, variety of emotions

becomes relevant when both positive and negative emotions have to be displayed

in contrast to having to display only positive emotions. For the samples analysed

in this article, it can be excluded that some of our participants are only required to

show negative emotions in their jobs. Rather, practically all participants had to

show positive emotions, which is demonstrated by the higher means of the

positive emotions scales compared to the negative emotions scales. If the display

of positive emotions is the starting point, then the requirement to display a high

variety of emotions means to have to display negative emotions as well.

Therefore, this scale can both be interpreted as the display of negative emotions

or the display of a high variety of emotions. The data also show that the

requirement to show a high variety of emotions does not guarantee that they can

more easily be matched than more homogeneous display requirements, as

implied by Morris and Feldman (1996). Rather, this scale obviously refers to

three problems: First, it seems to be more difficult to feel the emotion if a high

396

ZAPF ET AL.

variety of emotion display is required than if only one type of emotion should be

shown. Second, adaptation processes to show the various emotions may be

effortful, and third and most likely, in most cases the display of negative

emotions may be negative per se because they often indicate unpleasant social

interactions.

Against expectations, it was not possible to develop an emotion work control

scale. Actually, it was possible to develop scales with sufficient internal

consistency in the various samples, but the scale lacked discriminant validity.

The items for this scale also loaded on the interaction control factor and on the

emotional dissonance factor. In further research the item wording should be

improved to make the scale methodologically more sound. However, there

remains the problem of theoretical overlap, which makes achieving discriminant

validity difficult. This is necessarily so between interaction control and emotion

work control because they both were conceptualized to be sub-concepts of the

general control concept. However, the main problem lies in the discriminant

validity between emotion work control and emotional dissonance. Emotional

dissonance, as do all job stressors, implies a minimum of non-control. Otherwise

the stressor would simply be avoided. These conceptual reasons may make it

difficult to develop a scale with sufficient discriminant validity.

As in most of the other studies, it was possible to develop measures for

emotional dissonance with good scale properties and the expected correlations

with other emotion work and health variables. However, further research should

investigate the conceptualization of emotional dissonance as an objective

stressor, which could be supported by observation and peer evaluation.

The data suggest that the analysis of emotion work is a neglected area in

organizational stress research which should be given more attention in the future.

Based on the data, it can be suggested that emotion work is not per se either

positive or negative. Rather, emotion display and sensitivity requirements are

related to emotional exhaustion but also to personal accomplishment. In line with

Hochschild’s (1983) qualitative findings, it is emotional dissonance that is the

mismatch between the emotions that have to be displayed and the emotions that

one would like to display in a certain situation that is a stressor and that can lead

to psychological strain in the long run. Finally, it can be concluded that in jobs

where interacting with clients is a substantial part of the work, traditional

concepts of job stress do not suffice but should be complemented by concepts

measuring emotional requirements and emotional dissonance at work.

REFERENCES

Abraham, R. (1998). Emotional dissonance in organizations: Antecedents, consequences and

moderators. Genetic, Social, and General Psychology Monogaphs, 124, 229–246.

Adelmann, P.K. (1995). Emotional labor as a potential source of job stress. In S.L. Sauter & L.R.

Murphy (Eds.), Organizational risk factors for job stress (pp. 371–381). Washington, DC:

American Psychological Association.

EMOTION WORK

397

Ashforth, B.E., & Humphrey, R.H. (1993). Emotional labour in service roles: The influence of

identity. Academy of Management Review, 18, 88–115.

Basch, J., & Fisher, C.D. (1998, August). Affective events–emotion matrix: A classification of

work events and associated emotions. Paper presented at the first conference on Emotions and

Organizational Life, San Diego, CA.

Best, R.G., Downey, R.G., & Jones, R.G. (1997). Job burnout: A dysfunctional consequence of

contextual performance. Paper presented at the Convention of the Sociely for Industrial and

Organizational Psychology.

Bierhoff, H.W. (1990). Psychologie hilfreichen Verhaltens [Psychology of helpful behaviour].

Stuttgart, Germany: Kohlhammer.

Briner, R.B. (1995). Emotional dissonance, emotional deviance, true feelings, and the self in

organizational life. Paper prepared for the 12th EGOS Colloquium, Istanbul, Turkey.

Brotheridge, C.M., & Lee, R.T. (1998, August). On the dimensionality of emotional labor:

Development and validation of an emotional labor scale. Paper presented at the first

conference on Emotions in Organizational Life, San Diego.

Brucks, U. (1998). Arbeitspsychologie personbezogener Dienstleistungen [The work psychology

of person-related sevice]. Bern, Switzerland: Huber.

Bruggemann, A. (1974). Zur Unterscheidung verschiedener Formen von Arbeitszufriedenheit

[On the differentiation of various forms of job satisfaction]. Arbeit und Leistung, 28, 281–284.

Büssing, A., & Perrar, K.-M. (1992). Die Messung von Burnout. Untersuchung einer deutschen

Fassung des Maslach Burnout Inventory (MBI-D) [The measurement of burnout: Studies on a

German version of the Maslach Burnout Inventory]. Diagnostica , 38, 328–353 .

Cohen, S., & Wills, T.A. (1985). Stress, social support, and the buffering hypothesis.

Psychological Bulletin, 98, 310–357.

Cordes, C.L., & Dougherty, T. W. (1993). A review and integration of research on job burnout.

Academy of Management Review, 18, 621–656.

Doucet, L. (1998, August). Responsiveness: Emotion and information dynamics in service

interactions . Paper presented at the first conference on Emotions in Organizational Life, San

Diego.

Edwards, J.E., & van Harrison, R. (1993). Job demands and worker health: Three-dimensional

reexamination of the relationship between person–environment fit and strain. Journal of

Applied Psychology, 78, 628–648.

Ekman, P. (1984). Expression and the nature of emotion. In K.R. Scherer & P. Ekman (Eds.),

Approaches of emotion (pp. 319–344). Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

Ekman, P., & Davidson, R.J. (Eds). (1994). The nature of emotion: Fundamental questions. New

York: Oxford University Press.

Ekman, P., & Friesen, W.V. (1982). Felt, false and miserable smiles. Journal of Nonverbal

Behavior, 6, 238–252.

Erickson, R.J. (1991). When emotion is the product: Self, society, and inauthenticity in a

postmodern world. Unpublished doctoral dissertation, Washington State University, Pullman,

Washington DC.

Frese, M. (1987). A theory of control and complexity: Implications for software design and

integration of the computer system into the workplace. In M. Frese, E. Ulich, & W. Dzida

(Eds.), Psychological issues of human computer intercation in the work place (pp. 313–338).

Amsterdam: North-Holland.

Frese, M., & Zapf, D. (1987). Eine Skala zur Erfassung von Sozialen Stressoren am

Arbeitsplatz [A scale measuring social stressors at work]. Zeitschrift für Arbeitswissenschaft,

41, 134–141.

Frese, M., & Zapf, D. (1988). Methodological issues in the study of work stress: Objective vs.

subjective measurement of work stress and the question of longitudinal studies. In C.L. Cooper

& R. Payne (Eds.), Causes, coping, and consequences of stress at work (pp. 375–411).

Chichester, UK: Wiley.

398

ZAPF ET AL.

Frese, M., & Zapf, D. (1994). Action as the core of work psychology; A German approach. In

H.C. Triandis, M.D. Dunnette, & L.M. Hough (Eds.), Handbook of industrial and

organizational psychology (Vol. 4, pp. 271–340). Palo Alto, CA: Consulting Psychologists

Press.

Goffman, E. (1959). The presentation of self in everyday life. New York: Doubleday Anchor.

Goleman, D. (1995). Emotional intelligence: Why it can matter more than IQ. New York: Bantam

Books.

Goleman, D. (1998). Working with emotional intelligence. New York: Bantam Books.

Grandey, A. (1998, August). Emotional labor: A concept and its correlates. Paper presented at

the first conference on Emotions in Organizational Life, San Diego.

Greiner, B., & Leitner, K. (1989). Assessment of job stress: The RHIA-instrument. In K. Landau

& W. Rohmert (Eds.), Recent developments in job analysis: Proceedings of the international

symposium on Job Analysis, University of Hohenheim, March 14–15 1989 (pp. 53–66).

London: Taylor & Francis.

Hacker, W. (1973). Allgemeine Arbeits-und Ingenieurspsychologie [General work and

engineering psychology]. Berlin, Germany: Deutscher Verlag der Wissenschaften.

Hacker, W. (1998). Allgemeine Arbeitspsychologie. Psychische Regulation von Arbeits

tätigkeiten [General work psychology: Psychic regulation of work actions]. (4th ed.) Bern,

Switzerland: Huber.

Hackman, J.R. (1970). Tasks and task performance in research on stress. In J.E. McGrath (Ed.),

Social and psychological factors in stress (pp. 202–237). New York: Holt, Rinehart &

Winston.

Hill, C.A. (1987). Affiliation motivation: People who need people but in different ways. Journal

of Personality and Social Psychology, 52, 1008–1018.

Hochschild, A. (1983). The managed heart. Berkeley, CA: University of California Press.

Isic, A., Dormann, C., & Zapf, D. (in press). Belastungen und Ressourcen an Call Center-

Arbeitsplätzen [Stressors and resources in call-centres]. Zeitschrift für Arbeitswissenschaft.

Izard, C.E. (1977). Human emotions. New York: Plenum Press.

Jöreskog, K.G., & Sörbom, D. (1993). LISREL 8 (Computer program). Chicago: Scientific

Software.

Kahn, R., Wolfe, D., Quinn, R., Snoek, J., & Rosenthal, R. (1964). Organizational stress: Studies

in role conflict and ambiguity. New York: Wiley.

Kahn, R.L., & Byosiere, P. (1992). Stress in organizations. In M.D. Dunnette & L.M. Hough

(Eds.), Handbook of industrial and organizational psychology (Vol. 3, 2nd ed, pp. 571–650).

Palo Alto, CA: Consulting Psychologists Press.

Kruml, S.M., & Geddes, D. (1998, August). Catching fire without burning out: Is there an ideal

way to perform emotional labor? Paper presented at the first conference on Emotions in

Organizational Life, San Diego.

Lazarus, R.S., & Folkman, S. (1984). Stress, appraisal and coping. New York: Springer.

Lee, R.T., & Ashforth, B.E. (1996). A meta-analytic examination of the correlates of the three

dimensions of job burnout. Journal of Applied Psychology, 81, 123–133.

Marx, K. (l977). Capital (Vol. 1). New York: Vintage. (Original work published 1867).

Maslach, C. (1982). Burnout: The cost of caring. Englewood Cliffs, NJ: Prentice-Hall.

Maslach, C., & Jackson, S.E. (1986). Maslach Burnout Inventory (2nd ed.). Palo Alto, CA:

Consulting Psychologists Press.

Maslach, C., & Leiter, M. (1997). The truth about burnout. San Francisco: Jossey-Bass.

Mohr, G. (1986). Die Erfassung psychischer Befindensbeeinträchtigungen bei Arbeitern [The

measurement of psychological dysfunctioning of workers]. Frankfurt a.M. Peter Lang.

Mohr, G. (1991). Fünf Subkonstrukte psychischer Befindensbeeinträchtigungen bei Industriear-

beitern: Auswahl und Entwicklung [Five subconstructs of psychological dysfunctioning of

EMOTION WORK

399

industrial workers: Selection and development]. In S. Greif, E. Bamberg, & N.K. Semmer

(Eds.), Psychischer Streß am Arbeitsplatz (pp. 91–119). Göttingen, Germay: Hogrefe.

Morris, J.A. & Feldman, D.C. (1996). The dimensions, antecedents, and consequences of

emotional labor. Academy of Management Journal, 21, 989–1010.

Morris, J.A., & Feldman, D.C. (1997). Managing emotions in the workplace. Journal of

Managerial Issues, 9, 257–274.

Nerdinger, F.W. (1994). Zur Psychologie der Dienstleistung [The psychology of service].

Stuttgart, Germany: Schäffer-Poeschel.

Rafaeli, A. (1989). When cashiers meet customers: An analysis of the role of supermarket

cashiers. Academy of Management Journal, 32, 245–273.

Rafaeli, A., & Sutton, R.I. (1987). Expression of emotion as part of the work role. Academy of

Management Review, 12, 23–37.

Riggio, R.E. (1986). Assessment of basic social skills. Journal of Personality and Social

Psychology, 48, 649–660.

Rogers, C.R. (1951). Client-centered therapy. Boston: Houghton Mifflin.

Schaufeli, W.B. & Buunk, B.P. (1996). Professional burnout. In M.J. Schabracq, J.A.M.

Winnubst, & C.L. Cooper (Eds.), Handbook of work and health psychology (pp. 31–46).

Chichester, UK: Wiley.

Schaufeli, W., & Enzmann, D. (1998). The burnout companion to study and practice: A critical

analysis. London: Taylor & Francis.

Schaufeli, W., Maslach, C., & Marek, T. (Eds.). (1993). Professional burnout: recent

developments in theory and research. New York: Taylor & Francis.

Scherer, K.R. (1997). Emotion. In W. Stroebe, M. Hewstone, & G.M. Stephenson (Eds.),

Sozialpsychologie. Eine Einführung (3rd ed., pp. 293–330). Berlin, Germany: Springer.

Scherer, K.R., & Wallbott, H. (1990). Ausdruck von Emotionen [Expression of emotions]. In

K.R. Scherer, Enzyklopädie der Psychologie, Psychologie der Emotion (Vol. C/IV/3, pp. 345–

422). Göttingen, Germany: Hogrefe.

Semmer, N.K. (1984). Streßbezogene Tätigkeitsanalyse [Stress-oriented job analysis].

Weinheim, Basel, Switzerland: Beltz.

Semmer, N.K. (1996). Individual differences, work stress, and health. In M.J. Schabracq, J.A.

Winnubst, & C.L. Cooper (Eds.), Handbook of work and health psychology (pp. 51–86).

Chichester, UK: Wiley.

Semmer, N.K., & Baillod, J. (1991, August). Different forms of job satisfaction. Paper presented

at the congress of the swiss society of psychology, Lausanne, Switzerland.

Semmer, N.K., Zapf, D., & Dunckel, H. (1995). Assessing stress at work: A framework and an

instrument. In O. Svane & C. Johansen (Eds.), Work and health—scientific basis of progress in

the working environment (pp. 105–113). Luxembourg: Office for Official Publications of the

European Communities.

Semmer, N.K., Zapf, D., & Dunckel, H. (1999). Instrument zur Stressbezogenen

Tätigkeitsanalyse (ISTA) [Stress-oriented job analysis instrument ISTA) (pp. 176–204).

In H. Dunckel (Ed.), Handbuch zur Arbeitsanalyse. Zurich, Switzerland: Verlag der

Fachvereine.

Spector, P.E. (1987). Interactive effects of perceived control and job stressors on affective

reactions and health outcomes for clerical workers. Work and Stress, 1, 155–162.

Spector, P.E. (1992). A consideration of the validity and meaning of self report measures

of job conditions. In C.L. Cooper & I.T. Roberston (Eds.), International review of

industrial and organizational psychology, 1992 (Vol. 7, pp. 123–151). Chichester, UK:

Wiley.

Stenross, B., & Kleinman, S. (1989). The highs and lows of emotional labor. Journal of

Contemporary Ethnography, 17, 435–452.

400

ZAPF ET AL.

Strack, F., Stepper, L.L., & Martin, S. (1988). Inhibiting and facilitating conditions of the human

smile: A non-obstrusive test of the facial-feedback hypothesis. Journal of Personality and

Social Psychology, 54, 768–777.

Strauss, A., Farahaugh, S., Suczek, B., & Wiener, C. (1980). Gefühlsarbeit. Ein Beitrag

zur Arbeits- & Berufssoziologie [Sentimental work. A contribution to work & occupational

sociology].

Kölner

Zeitschrift für Soziologie und

Sozialpsychologie, 32,

629–651.

Strazdins, L. (1998). Integrating emotions: Multiple role measurement of emotional work.

Australian National University, Canberra, Department of Psychology.

Tolich, M.B. (1993). Alienating and liberating emotions at work: Supermarket clerks’

performance of customer service. Journal of Contemporary Ethnography, 22, 361–381.

Volpert, W. (1974). Handlungsstrukturanalyse als Beitrag zur Qualifikationsforschung [Action

structure analysis: A contribution to qualification research]. Köln, Germany: Pahl-Rugenstein.

Wall, T.D., Jackson, P.R., Mullarkey, S., & Parker, S.K. (1996). The demand–control model of

job strain: A more specific test. Journal of Occupational and Organizational Psychology, 69,

153–166.

Wanous, J.P., Reichers, A.E., & Hudy, M.J. (1997). Overall job satisfaction: How good are single

item measures. Journal of Applied Psychology, 82, 247–252.

Wiemann, J.M., & Giles, H. (1997). Interpersonale Kommunikation [Interpersonal com-

munication]. In W. Stroebe, M. Hewstone, & G.M. Stephenson (Eds.), Sozialpsychologie. Eine

Einführung (3rd ed., pp. 331–366). Berlin: Springer.

Wharton, A. (1993). The affective consequences of service work: Managing emotions on the job.

Work and Occupations, 20, 205–232.

Zapf, D. (1993). Stress-oriented job analysis of computerized office work. The European Work

and Organizational Psychologist, 3, 85–100.

Zapf, D., Dormann, C., & Frese, M. (1996). Longitudinal studies in organizational stress

research: A review of the literature with reference to methodological issues. Journal of

Occupational Health Psychology, 1, 145–169.