Antecedents of Workplace Emotional Labor

Embed Size (px)

Citation preview

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    1/21

    Antecedents of workplace emotional labordimensions and moderators of their eectson physical symptoms

    JOHN SCHAUBROECK*1 AND JAMES R. JONES2

    1Department of Management, LeBow College of Business, Drexel University, 33rd & Arch Streets,Philadelphia, PA 19104, U.S.A.

    2Department of Marketing/Management, College of Business Administration, University of Nebraskaat Omaha, Omaha, NE, U.S.A.

    Summary The present study distinguished between two modal emotional display rules, demands toexpress positive eerence and demands to suppress negative eerence, that partiallyconstitute the work roles of many employees. Perceived demands to express positiveemotion were positively related to health symptoms primarily among those reporting: (1)lower identication with the organization; (2) lower job involvement; and (3) loweremotional adaptability. The eects of various personality traits and situational variableson perceived emotional labor diered depending on the nature of the emotional labor.The ndings are discussed in terms of implications of emotional labor for health andpractices through which organizations might intervene to minimize its unhealthfulconsequences among employees. We also attempt to reconcile the ndings with some ofthe related research in psychology suggesting that some forms of required eerence mayhave salutary physiological consequences. Copyright # 2000 John Wiley & Sons, Ltd.

    Introduction

    Much research investigating the relationship between job demands and health has emphasized

    the key psychosocial nature of the demands. Pressures are placed on individuals in a social

    context and their behavioral responses to these pressures have socially derived implications for

    them. Being required to modulate the expression of one's own emotions in particular ways is a

    signicant component of the work role for many persons. This type of role requirement has been

    called `emotional labor' (cf. Wharton and Erickson, 1993, p. 458). The concept of emotionaldisplay rules was introduced by Paul Ekman (e.g., Ekman, 1972). Emotional display rules refer to

    norms about appropriate emotional expression for specic situations. Persons who have much

    customer or client contact (e.g., salespersons, nurses) are seen to be subject to stronger emotional

    display rules (Sutton, 1991; Sutton and Rafaeli, 1988). Several authors have observed that such

    display rules may be expected to compromise the psychological and/or physical health of workers

    CCC 08869383/2000/02016321$17.50Copyright # 2000 John Wiley & Sons, Ltd.

    Journal of Organizational Behavior

    J. Organiz. Behav. 21, 163183 (2000)

    * Correspondence to: John Schaubroeck, Department of Management, LeBow College of Business, Drexel University,33rd & Arch Streets, Philadelphia, PA 19104, U.S.A. E-mail: [email protected]

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    2/21

    because they often lead to a disturbing disequilibrium (or `dissonance') between felt emotions

    and the emotions one must exhibit (Ashforth and Humphrey, 1993; Morris and Feldman, 1996;

    Rafaeli and Sutton, 1987; Wharton and Erickson, 1993). Indeed there is considerable evidencefrom outside the work sphere that chronically experiencing conict among one's emotions has

    negative health consequences (see reviews by Friedman, 1989; King and Emmons, 1990).

    Emotional labor is a construct of increasing interest among theorists in the management

    literature (cf. Ashforth and Humphrey, 1993; Morris and Feldman, 1996; Rafaeli and Sutton,

    1987; Wharton and Erickson, 1993).

    From an empirical point of view, however, `the expression of socially appropriate emotion

    constitutes a neglected form of role demand' (Ashforth and Humphrey, 1993, p. 110). As noted

    by Wharton and Erickson (1993), `virtually all previous studies [of emotional labor] have been

    case studies' ( p. 462). To our knowledge, only two quantitative studies of emotional labor in the

    workplace have been published in the literature. After sampling workers in two companies,

    Wharton (1993) constructed a dummy variable of emotional labor by using a scheme developed

    by Hochschild (1983) to classify some of the jobs as high in emotional labor. Wharton (1993)

    observed no signicant correlation between this variable and emotional exhaustion, and a

    signicant negative correlation with job satisfaction. Morris and Feldman (1997) developed a

    scale measure of emotional labor dierentiated along three dimensions: duration of contact with

    others, frequency of contact with others, and emotional dissonance inspired by such contact.

    Their study supported hypotheses linking emotional labor to characteristics of the job as well as

    emotional exhaustion.

    Despite the claims of Hochschild (1983) and others that emotional labor is related to physical

    symptoms, there is no research relating emotional labor to physical symptoms. As we review

    below, laboratory studies from psychology are suggestive of the possibility that certain types of

    emotional labor may in fact have salutary inuences on individual's physiological well-being

    whereas other forms may be toxic to health. We distinguish in this study between demands to

    suppress negative emotion, which may generally be expected to be unhealthful, and demands to

    express positive emotion. Second, we identify key individual and situational moderators of thisemotional laborsymptoms relationship. Third, it is not clear whether or to what extent

    emotional labor reects dispositional tendencies and/or objective situational exposures. Accord-

    ingly, we examine how individual dierences and situational variables determined two dimen-

    sions of emotional labor and also how these variables moderated relationships between

    emotional labor and health symptoms.

    Figure 1 depicts the framework for this study. Objective characteristics of the role

    (i.e., requirements for interpersonal interaction) and personal traits (i.e., personality traits and

    gender) are posited to determine perceptions of emotional labor. Perceptions of emotional labor

    inuence physical health. The relationship between perceived emotional labor and health is

    moderated by emotional adaptability and work perceptions (i.e., job involvement and organiz-

    ational identication).

    Psychological research on emotional expression

    Dierent aspects of emotional labor have been explicated and many propositions about their

    eects on well-being and other outcomes have been proposed. However, the prevailing concept of

    emotional display rules remains unidimensional in character despite the range of plausible types

    of display rules. For the present study we instead chose to focus on the character of the emotions

    being managed, such as whether a particular type of eerence (i.e., positive or negative) is

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    164 J. SCHAUBROECK AND J. R. JONES

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    3/21

    required to be expressed or suppressed. Perhaps because the concept emerged from a dierent

    eld (sociology), research and theory about emotional labor has generally overlooked relevant

    research from the eld of psychology. However, the psychological literature may usefully

    broaden perspectives about emotional labor. In addition to the emotional conict research

    conducted in eld settings (as reviewed by King and Emmons, 1990), a good deal of laboratory

    research has investigated the relationship between emotional eerence and physiological

    outcomes. The psychological studies of emotion have tended to distinguish between two types

    of emotion manipulation: the suppression of felt negative emotion and the required display of

    positive emotion. While there are many possible types of display rules, we believe these twogeneral categories, the display of positive emotional eerence and the suppression of negative

    emotional eerence, capture many of the rules operating in organizational contexts. When one is

    explicitly or implicitly required to display emotional eerence in the organization, typically one is

    required to display cheerfulness, earnestness, and other emotions that are culturally dened as

    `positive'. Exceptions to this may include psychological therapists, morticians, and other job-

    holders that are often required to `deep act' (Rafaeli and Sutton, 1987) sadness, concern, and

    other less `positive' emotions (cf. the bill collectors studied by Sutton, 1991). In addition,

    interpersonal role relationships such as one has with coworkers and customers often require one

    to suppress negative emotions such as irritation and anger. Whereas these two types of role

    requirements may not capture all forms of emotional labor, we believe they represent emotional

    labor demands that are relevant to a wide range of occupations.

    Occupational categorization of

    emotional labor

    The present study takes a psychological, rather than sociological, approach to emotional labor.

    Thus in contrast to previous studies we do not equate emotional labor with the characteristics of

    an occupational setting (cf. Hochschild, 1983) or the frequency and duration of interaction with

    other persons (Morris and Feldman, 1997). Rather we operationalize requirements to express or

    suppress emotion in terms of the experience of workers in having to suppress negative emotions

    Figure 1. Overview of research model

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    WORKPLACE EMOTIONAL LABOR DIMENSIONS 165

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    4/21

    and express positive emotions in order to meet the demands of their jobs. The `emotional

    dissonance' measure used by Morris and Feldman (1997) comes closest to our approach in

    measuring emotional labor (e.g., `Most of the time, the way I act and speak with patients matcheshow I feel anyway' (reverse-scored; p. 263)). This type of measure appears on its face to be more

    consistent with existing denitions of emotional labor than is mere interpersonal interaction.

    Whether being required to interact frequently with others creates emotional labor may be

    expected to depend on some interaction between the context of the interactions and the charac-

    teristics of the individual. Indeed, Morris and Feldman (1997) observed very little correlation

    between the frequency and duration of interactions with others and their index of emotional

    dissonance. As a sociologist, however, Hochschild (1983) was more interested in measuring the

    distribution of objective occupational dierences and exploring how incumbents of these

    occupations had to manage their emotions. It should also be noted that Hochschild's scheme for

    identifying occupations that require much emotional labor did take the expected modal content

    of interactions into account, not just the frequency and duration of these. However, her coding

    scheme did not distinguish this content in any way, and like all broad occupation variables hers

    provides a crude estimate of the psychological exposures experienced by dierent individuals in

    the same occupation. As we examine further on in this paper, it may well be that individual

    dierences inuence the perception and experience of emotional labor. Thus one goal of this

    study was to ascertain the extent to which emotional labor, as measured directly at the individual

    level, corresponds with the occupational scheme developed by Hochschild that is based primarily

    upon the volume and intensity of interpersonal interaction on the job.

    Hypothesis 1. The extent to which individuals' occupations require interpersonal interaction

    will be positively related to their reports of role expectations to suppress negative emotional

    eerence and to display positive emotional eerence.

    The role of emotion-related traitsin emotional labor

    In addition to emotional labor inherent within a particular job or occupation, the present study

    also seeks to examine its dispositional sources. In the case of emotional labor, the key resource

    domain for coping may be related to self-perceptions of the emotional resources and capabilities

    that are specic to emotional experience. Two separate literatures, one in sociology and the

    other in psychology, suggest a critical individual dierence variable. A sociological construct of

    `institutional' versus `impulsive' orientation of emotions refers to the extent to which `people may

    nd evidence of their real, deeper selves either in institutional behavior and feelings, or else in

    impulsive acts and feelings' (Gordon, 1989, p. 117). Persons whose emotions are institutionally-

    oriented `focus [their] eorts on achieving and maintaining institutional standards for feeling and

    expression, thereby upholding widely shared values', whereas `a perspective centered on impulse

    [italics his] emphasizes spontaneous, uninhibited emotion, unregulated by institutions`. Ashforthand Humphrey (1993) suggested that persons having an `institutional' orientation to their

    emotions may be expected to experience less emotional dissonance and higher well-being in

    conforming to display rules than do persons having an `impulsive' orientation ( p. 100).

    Within the eld of psychology, Laird and Crosby (1974) examined trait dierences in the extent

    to which persons used `self-produced cues' versus `situationally produced cues' in making self-

    attributions about their own emotions. Participants were required to either frown or smile while

    watching various cartoons. Those who reported that their emotions were typically self-

    produced demonstrated a signicantly stronger relationship between the emotional expression

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    166 J. SCHAUBROECK AND J. R. JONES

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    5/21

    manipulation and the moods they experienced just after the cartoons. There are evident parallels

    between the `institutional orientation' described by Gordon (1989) and this proclivity for one's

    emotional expression to follow situation-produced cues, just as there are parallels betweenGordon's `impulsive orientation' and the tendency to use `self-produced cues' as stimuli for

    emotional expression. People who maintain an other-oriented disposition toward the expression

    of emotion (i.e., `institutional' or `situation-produced') are essentially more willing and able to

    adapt their emotional expression to the social environment. Hochschild (1983) referred to this as

    an ability to develop a `healthy false self' in which `inauthentic' emotions are readily expressed in

    service of the role ( p. 195). Hereafter we refer to this trait as `emotional adaptability'.

    These very similar lines of inquiry suggest that adaptability to emotional demands is an

    individual dierence variable. Although emotional adaptability may be construed in numerous

    ways, for the present purposes we dene persons who naturally adapt their emotions to the

    situational context as being `emotionally adaptable'. Persons who are more emotionally adapt-

    able in this respect may tend to report fewer requirements to express or suppress emotion. This is

    because one's perception of the normative requirements to express or suppress emotion is at least

    partially an attribution that is sensitive to self-perception eects. Specically, when one readily

    meets demands for emotional expression, he or she does not require explicit cues from other co-

    workers or the supervisor to emote appropriately. Thus expressions of positive emotion are not

    easily attributed to role obligations. On the other hand, persons who must put more eort into

    appropriate emotional expression or who nd this aspect of their role dicult will be made more

    cognizant of social cues about norms for appropriate emotional expression. Moreover, the very

    challenge of this aspect of the role will make appropriate emotional expression more easily

    recognized as a role requirement for these persons than it is for emotionally adaptable persons.

    Thus we predict that whereas more emotionally adaptive persons may or may not emote

    dierently compared to less emotionally adaptive persons, the former will not experience role

    appropriate emotional expression as emotional labor to the same extent as the latter.

    Hypothesis 2. Emotional adaptability will be negatively related to reports of requirements toexpress or suppress emotion.

    When researchers speak of a dissonance created by demands to display emotions that one does

    not actually feel, then emotional labor must be seen as an interaction between dispositions of the

    individual and his or her job context. Persons are more likely to nd their spontaneous emotions

    to be at odds with emotional display rules when they have particular aect tendencies. Research

    has identied two basic, essentially orthogonal trait continua pertaining to aect. Trait positive

    aect (PA) refers to a tendency for one to be frequently energetic, active, alert, and enthusiastic.

    Persons scoring high on trait negative aect (NA) tend, even in normal situational contexts, to

    more frequently experience negative emotions such as anxiety, disgust, guilt, and fear. Trait PA is

    associated with positive evaluations of social environments, whereas trait NA is associated with

    correspondingly negative evaluations (Watson and Clark, 1984; Watson et al., 1988). A person'saect tendencies may lead directly to higher perceptions of requirements to express or suppress

    emotion because particular aective styles will be more or less congruent with particular

    emotional display rules. A person who naturally experiences positive emotions on a regular basis,

    for example, may hardly be cognizant of others' expectations to `act cheerful'. Such behavior is

    not likely to be perceived as role-oriented for such a person, but rather authentic. Likewise, a high

    NA person may have diculty suppressing the expression of negative emotions since these are

    felt more frequently and deeply than for a low NA person. Thus any social cues that he or she

    receives concerning the unacceptability of negative eerence (cues to which `cheerful' people are

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    WORKPLACE EMOTIONAL LABOR DIMENSIONS 167

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    6/21

    seldom exposed) may be perceived to dene emotional display rules for such persons. This

    suggests the following hypotheses:

    Hypothesis 3. Trait positive aect will be negatively associated with perceptions of role

    expectations to display positive emotional eerence and negatively associated with role

    expectations to suppress negative emotional eerence.

    Hypothesis 4. Trait negative aect will be positively associated with perceptions of role

    expectations to suppress negative emotional eerence and positively associated with role

    expectations to express positive emotional eerence.

    Trait aect reects one's modal aective states and thus represents states experienced on the

    job as well as o the job. For some persons modal state aect may dier between work and other

    domains, and their aect on the job may more strongly inuence perceived emotional labor.

    Therefore, in conjunction with examining the eects of trait aect on perceived emotional labor,

    we also examined the eects of state aect while the individual is at work.

    Gender and emotional labor

    Hochschild (1983) described the close relationship between emotional labor and stereotypical

    gender roles in American society. `In general, they [women] are thought to manage expression

    and feeling not only better but more often than men do . . . the evidence seems clear that women

    do more emotion managing than men' (both italics hers; p. 164). Indeed, much of the literature on

    emotional labor focuses on women's roles, both at work and at home. Hochschild (1983) did not

    distinguish conceptually between the two dimensions of requirements to express or suppress

    emotion proered in this study. However, in her studies of female ight attendants, Hochschild

    referred to both facets in referencing the higher prevalence of emotional labor among women:`Women . . . do extra emotion workespecially emotion work that arms, enhances, and

    celebrates the well-being and status of others' [italics hers]. She also suggested that `Women are

    more likely to be presented with the task of mastering anger and aggression in the service of

    ``being nice'' ' (1983, p. 163). These suppositions allude to the distinction we have made between

    demands to express positive emotion and demands to suppress negative eerence. Whereas

    Wharton (1993) found a higher prevalence of women in jobs that tend to require high amounts of

    emotional labor (using Hochschild's occupational coding scheme), that relationship was

    explained by other demographic variables and income level. But Hochschild's discourse about

    women and emotional labor goes beyond the sociological task of describing pre-market factors

    which dierentially assign the sexes to occupations. Rather, her discussion of the social

    psychology of women in all manner of occupations, including traditionally male roles, suggests

    that women will experience more emotional labor even after the sociological status factors aretaken into account. Women were suggested to take more initiative in undertaking emotional

    labor at work, and co-workers were suggested to have higher expectations of emotion work from

    women: `The world turns to women for mothering, and this fact silently attaches itself to many a

    job description' (Hochschild, 1983; p. 170). This suggests the following hypothesis:

    Hypothesis 5. After controlling for pre-market factors that may dierentiate women and men

    generally, women will still be found to report higher levels of requirements to express or

    suppress emotion than men.

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    168 J. SCHAUBROECK AND J. R. JONES

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    7/21

    Direct and interactive eects on physical symptoms

    Laboratory research suggests that requirements to display emotions often lead to physical andemotional reactions that are consistent with the instructions. As stated by Notarious et al. (1982),

    `Facial expressions are subject to strong personal monitoring and voluntary control and thus

    likely vary as a function of the display rules operating in a given context' ( p. 401). One related line

    of laboratory research in psychology concerns the `facial feedback hypothesis' (Izard, 1971). As

    reviewed by Adelmann and Zajonc (1989), facial feedback studies typically require participants

    to manipulate their facial expressions in ways that contradict the emotional context of their

    immediate environment. Negative emotional eerence (e.g., frowning) is found to promote

    adverse emotional experience and physiological indicators, just as positive emotional eerence

    (e.g., smiling) promotes positive physical and psychological reactions. This research suggests that

    positive emotional display rules (e.g., `always smile when one is on the job') might have salutary

    inuences on health.

    Hypothesis 6. Requirements to express positive emotional eerence on the job will benegatively associated with physical symptoms.

    Laboratory ndings indicate that eorts to suppress `negative' emotions (i.e., anger,

    irritability, grief, sadness) often lead to patterns of physiological response that presage somatic

    illness. When one is prevented from expressing emotions in a psychologically stimulating context,

    generally a `discharge model' of emotion is supported, wherein high levels of physiological

    reactivity and negative moods are experienced (e.g., Notarious and Levenson, 1979; Notarious

    et al., 1982). Likewise, the psychological characteristics of denial and suppression are consistently

    linked to lower immune levels and susceptibility to viral infection (see the review by S. Cohen and

    T. B. Herbert, 1996) as well as cardiovascular illness (see the review by Friedman, 1989). Thus

    this type of emotional labor is expected to be positively associated with physical symptoms.

    Hypothesis 7. Requirements to suppress negative emotional eerence on the job will bepositively associated with physical symptoms.

    Role identication moderators

    Rationalization processes may buer individuals from the health eects of requirements to

    express or suppress emotion. Specically, individuals who identify strongly with their organiza-

    tions and/or their jobs may more fully subscribe to the belief that they must often behave in an

    emotionally `inauthentic' fashion to serve the purposes of their roles. Thus they will experience

    little cognitive dissonance following such `inauthentic' acts. Varied research indicates that,

    provided that they are fully identied with a higher organizational purpose, people have little

    trouble behaving in ways they would normally nd unnatural or even abhorrent. When anindividual lacks that sense of purpose, however, considerable emotional dissonance may be

    created by his or her inauthentic emotional behavior.

    Hypothesis 8. Perceived requirements to express or suppress emotion will interact with

    organizational identication and job involvement to predict physical symptoms. As

    organizational identication or job involvement increase, there will be signicantly weaker

    relationships between requirements to express or suppress emotion dimensions and physical

    symptoms.

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    WORKPLACE EMOTIONAL LABOR DIMENSIONS 169

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    8/21

    Individual dierence moderators

    Individual dierences in personality may not only inuence the extent to which people perceive

    role requirements for emotional expression as noted above, they may also inuence the strengthof the relationship between these perceptions and physical symptoms. As noted above, the less

    emotionally adaptable the individual is, the more he nds it dicult and/or onerous to modulate

    his emotions in ways that are socially appropriate. Thus the emotionally adaptive individual may

    be better able to cope with perceived requirements for emotional expression than is the less

    emotionally adaptive person.

    Hypothesis 9. As emotional adaptability increases, requirements to express or suppress

    emotion will be less strongly related to physical symptoms.

    Methods

    Sample and procedure

    Questionnaires were collected from the full-time headquarters employees of a major survey

    research organization in the mid-west United States. Because the nature of much of the work in

    this organization is rather complex and requires internal and external interactions with people,

    the site provided a fairly high incidence of emotional labor. A total of 227 persons returned a

    questionnaire, of which 217 included complete data for the analyses.

    Nearly all occupations within the organization were represented in the sample. The average

    respondent's age was 36.84 years, and 62 per cent of participants were women. Respondents

    averaged 7.31 years in the organization and 4.29 years in their current position. The mean level of

    educational attainment was quite high, averaging 0.58 years of postgraduate college education.

    Respondents did not dier signicantly from the overall organization on any of these

    demographic data.

    Measures

    Emotional labor

    Our approach to measuring emotional labor was to index the character of the emotions that are

    perceived to be encouraged on the job. The items were scaled using a ve-point Likert format.

    They focused on norms that must be complied with for: (1) eective job performance; or (2) `To

    make a good impression on others (e.g., bosses, co-workers, customers, etc.) in my job'.

    We formed a composite of demands for positive eerence (a 0.87) and suppression of

    negative eerence (a

    0.96) using the products of duration and frequency items. This type ofcomposite measure therefore reects the degree of emotional labor by combining the frequency

    and the overall amount of attention devoted to these demands. This was done because it is

    unlikely, for example, that someone who believes it is a norm to smile at passers-by when walking

    across an oce space experiences this as demanding. Likewise, a worker who has only one or two

    emotional display rule-laden interactions per day may nd this demanding if these encounters are

    lengthy.

    All items were subjected to a principal factor analysis with oblique rotation. All item-products

    loaded distinctly with coecients exceeding 0.40 on one of two factors, both with eigenvalues

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    170 J. SCHAUBROECK AND J. R. JONES

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    9/21

    exceeding 1.0 (6.87 and 3.35, respectively). The rst factor comprised the positive eerence items

    (e.g. `To be eective in my job, I must try to share in the enthusiasm or liveliness of another

    person.') and explained 42.9 per cent of the overall item variance. The item-products loading onthe second factor measured suppression of negative eerence (e.g., `To be eective in my job, I

    must try to suppress how upset or distressed I may feel.'). These item-products explained the next

    21.0 per cent of the variance. (The complete list of emotional labor items is available from the

    rst author upon request.)

    Interpersonal interactionoccupation

    In order to further guarantee condentiality, rather than asking for the respondent's specic job

    title the questionnaire inquired about his or her job category from a list of 35 occupational units

    present in the organization. Most all of these categories were readily related to the broad

    occupational dimensions which Hochschild (1983) identied as having high levels of emotional

    labor. We also asked two managers to identify which categories on the list contained high needs

    for intensive interaction with others. A few job categories were not in Hochschild's list ofoccupations but clearly required high levels of intensely personal interaction with other persons

    (i.e., Executive Interviewer, Research Consultant, Interviewer, Computing Services operational

    support). Given the service nature of the organization, it is not surprising that a signicant

    portion of the respondents who identied their occupational category (n 195) were included in

    the `high emotional labor' group (72.3 per cent). The remaining occupational categories were not

    on Hochschild's list. The managers indicated that these latter occupations required little inter-

    personal interaction or those interactions that were required were not qualitatively demanding in

    interpersonal interaction. Examples of jobs that were not coded as being high in emotional labor

    were building services, research analyst, data coding/entry, mailing/scanning/proong, and

    accountant. High emotional labor job categories were coded `2' and the other categories were

    coded `1' to form an interpersonal interaction variable.

    Physical symptoms

    Physical symptoms were measured using Caplan et al.'s (1975) 20-item somatic complaints index

    (a 0.86). Respondents are asked `How often have you experienced any of the following events

    during the past month? Rarely (1) to Very often (5)', (e.g., `You were in ill health that aected

    your work'; `You had trouble sleeping at night.`).

    Individual dierence variables

    Age, gender, trait negative aect, and education were measured on the questionnaire. These were

    included as control variables in tests of models that examined physical symptoms as a dependent

    variable. We also included quantitative measures of smoking and physical exercise, but these

    variables did not correlate signicantly with physical symptoms. State negative and positive aectwere controlled in analyses examining trait aect and its correlations with requirements to express

    or suppress emotion. Trait negative aect has been found to be a confound variable in examining

    relationships between stressors and health outcomes when these are both measured by self-

    reports (Brief et al., 1988).

    For the analyses examining the relationship between gender and perceived requirements to

    express or suppress emotion, we also controlled for self-reported salary and job complexity. Job

    complexity was measured using a shortened six-item version of the instrument used by Caplan

    et al. (1975) and House (1980). These items (e.g., `How often does your job require that you think

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    WORKPLACE EMOTIONAL LABOR DIMENSIONS 171

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    10/21

    through complex problems or situations?') provided an index of the extent to which the job was a

    traditionally `male-gendered' role carrying higher social status, and thus together with the other

    controls (e.g., education) it assisted in controlling for social status. Salary data were obtained by aseparate survey designed by the organization and conducted one month earlier. Data on salary

    and the other variables combined were thus available for just over one-half (n 123) of the

    respondents. The respondent was asked to indicate which of six salary range categories included

    his or her own salary.

    We used the Positive and Negative Aect Scale (PANAS; Watson et al., 1988) to measure both

    forms of aect. There are 20 adjectives, 10 reecting negative emotional states (e.g., `irritable`)

    and 10 reecting positive emotional states (e.g., `excited'). The PANAS respondent is asked to

    record the extent (from `very slightly or not at all' [1] to `extremely' [5]) to which he or she has

    experienced the emotion during a specied interval. Watson et al. (1988) note that this interval

    can be dened to represent state aect ranging from momentary states, single days, weeks, to

    years, with `generally feel that way' representing trait aect. Both the positive state aect (PA)

    and negative state aect (NA) scales evidenced sound internal consistency reliability (a 0.91

    and 0.83, respectively). Watson et al. (1988) observed similarly strong internal consistencies for

    `momentary' PANAS scales as well as reasonably high testretest reliabilities over an eight week

    period (r 0.54 (PA) and r 0.45 (NA)). For our state aect indexes, `during the past hour' was

    the reference period.

    The same PANAS questionnaire modied to measure trait aect (`. . . indicate the extent you

    generally feel that way [both at work and away from work], that is, on average.') was included

    near the end of the questionnaire. The trait PA and the trait NA items provided coecient alpha

    reliabilities of 0.89 and 0.87, respectively.

    'Emotional adaptability' represents the extent to which one's expressed emotions are adapted

    easily to the institutional context. This instrument was adapted from three items developed by

    Laird and Crosby (1974); (a 0.68; e.g., `Frequently one moves from a situation on the job in

    which one is feeling one way to a situation in which it is appropriate to feel some other way, as,

    for instance, when you have an argument with someone and then immediately must be withsomeone else. How rapidly do you change from the rst kind of feeling to the second? 1 very

    slowly to 4 very rapidly').

    Other dispositional variables

    We adapted the ve items developed by Mael (1988unpublished doctoral dissertation, Wayne

    State University, Detroit) to measure organizational identication. The items (a 0.92) were

    adapted by making specic reference to the name of the organization (e.g., `When someone

    criticizes [name of organization], it feels like a personal insult.').

    The three-item job involvement instrument from the Michigan Organizational Assessment

    Questionnaire (MOAQ; Camman et al. (1969)) used a 17 `very inaccurate' to `very accurate'

    scale (a 0.84; `I am very much involved personally in my job'; `I live, eat, and breathe my job';

    `The most important things which happen to me involve my job').

    Results

    Correlations among all variables are shown in Table 1. Emotional adaptability was positively

    associated with state and trait positive aect and negatively correlated with state and trait

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    172 J. SCHAUBROECK AND J. R. JONES

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    11/21

    Table 1. Correlations among study variables and descriptive statistics

    Variable name Mean S.D. 1 2 3 4 5 6 7 8 9 10 11 12

    1. Age (years) 36.84 12.32. Gender (1 male; 2 female) 1.62 0.49 0.063. Education (years) 16.58 13.1 0.22 0.23 4. Tenure ( job) 4.29 4.91 0.44 0.13 0.19 5. Job complexity 3.80 0.79 0.19 0.03 0.29 0.17 6. Demands to express positive emotions 11.33 5.83 0.0 8 0.2 2 0 .07 0.03 0.10 7. Demands to suppress negative emotions 5.53 4.39 0.1 2 0.1 1 0 .03 0.11 0.04 0.35 8. Physical symptoms 1.85 0.53 0.27 0.30 0.08 0.21 0.07 0.28 0.43

    9. Organization identication 5.89 1.32 0.16 0.16 0.03 0 .00 0.2 3 0.1 4 0.23 0.03 10. Job involvement 4.96 1.45 0.21 0.01 0.21 0.09 0.41 0.13 0.18 0.02 0.67 11. Emotional adaptability 2.74 0.83 0.11 0.12 0.16 0 .01 0.1 5 0.0 7 0.04 0.04 0.25 0.17 12. State positive aect 3.33 0.86 0.18 0.11 0.22 0.05 0.27 0.08 0.21 0.14 0.36 0.44 0.24 13. Trait positive aect 3.93 0.62 0.21 0.09 0.20 0.06 0.36 0.09 0.31 0.20 0.39 0.44 0.23 0.5814. State negative aect 1.55 0.55 0.1 1 0.0 9 0 .09 0.09 0.02 0.13 0.37 0.42 0.17 0.13 0.07 0.1815. Trait negative aect 1.76 0.60 0.90 0.12 0.05 0.09 0.1 6 0.0 5 0. 48 0. 58 0.25 0.17 0.17 0.2116. Emotional labor occupation (Hochschild, 1983) 1.72 0.45 0.12 0.06 0.02 0.12 0.16 0.17 0. 03 0. 04 0 .0 1 0.01 0.03 0.0417. Salary 3.49 1.69 0.27 0.21 0.39 0.30 0.33 0.17 0.07 0.05 0.01 0.24 0.0 2 0. 23

    Variables 115: pairwise n's 218222 (critical value of r (p 5 0.05) 0.13; p 5 0.01 0.18; p 5 0.001 0.23).Variable 16: pairwise n's 191195 (critical value of r (p 5 0.05) 0.15; p 5 0.01 0.19).Variable 17: pairwise n's 121123 (critical value of r (p 5 0.05) 0.19; p 5 0.01 0.24; p 5 0.001 0.30).

    Copyright#

    2000JohnWiley&Sons,Ltd.

    J.Organiz.Behav.21,163183(2000)

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    12/21

    negative aect. The two measures of emotional labor, demands to suppress negative eerence and

    demands to express positive eerence, correlated signicantly (r 0.35, p 5 0.0001).

    Predictors of emotional demands

    Consistent with Hypothesis 1, Hochschild's (1983) coding of the occupational units in this

    sample for `objective' expectation of emotional labor (based largely on expected quantity of

    interpersonal interaction) was positively correlated with demands to express positive emotion

    (r 0.17, p 5 0.01). However, the hypothesis was only partially supported as this variable did

    not correlate with demands to suppress negative eerence (r 0.03).

    The second hypothesis predicted that emotional adaptability would be negatively associated

    with reports of requirements to express or suppress emotion. After controlling for trait negative

    aect, age, education, and gender, emotional adaptability was not associated with perceived

    demands to suppress negative emotion (DR2 0.00, F(1, 206) 0.72, p 5 0.40), nor was this

    trait associated with perceived demands to express positive emotion on the job ( DR2 0.01, F(1,

    206) 2.31, p 5 0.13). Thus Hypothesis 2 was not supported.

    To examine the incremental contributions of the dispositional aect variables on emotional

    labor perceptions, we rst regressed each emotional labor index on the demographic controls

    (age, sex, and education) and then we entered both trait aect variables. These variables ( trait

    NA and PA) together explained a good deal of the variance in demands for suppression of

    negative eerence (DR2 0.25, F(4, 203) 18.09, p 5 0.0001), but very little of the variance in

    demands for positive emotional expression (DR2 0.01, F(4, 203) 0.56, p 5 0.67). Thus

    Hypothesis 3 (pertaining to trait PA) and Hypothesis 4 (pertaining to trait NA) were supported

    for the suppression of the negative eerence variable but not the positive emotional expression

    variable. The state aect variables (state PA and NA) next entered the equation. After controlling

    for trait aect, the block of state aect variables was marginally related to demands forsuppression of negative eerence ((DR2 0.01, F(2, 203) 2.85, p 5 0.06) but it was unrelated

    to demands for positive expression (see Table 2).

    Gender ( female 2, male 1) correlated positively with perceived demands to express

    positive emotion (r 0.22, p 5 0.001). There was no relationship between gender and

    Hochschild's (1983) coding of occupations that are seen as high in emotional labor (`interpersonal

    interactionoccupation'; r 0.06, n.s.). To test Hypothesis 5, positive eerence demands was

    rst regressed on age, education, job complexity, position tenure, trait negative aect, trait positive

    aect, and salary level (all employees were salaried). As noted above, salary data were available for

    only half of the respondents. Together, these variables explained 12 per cent of the variance in

    positive eerence demands (F(7, 111) 2.13, p 5 0.05). Gender was entered at the next step, and

    its strong incremental eect on positive eerence demands (DR2 0.06, F(1, 110) 8.18,

    p 5 0.005) supported the hypothesis (see Table 2). Gender was not signicantly related toperceived demands to suppress negative eerence.

    Hypothesis 6 predicted a negative relationship between reported demands for positive eerence

    and physical symptoms. This hypothesis was not supported as there was a positive eect of these

    kinds of demands on symptoms, after controlling for trait negative aect, age, education, and

    gender (see Table 3). The eect of demands to suppress negative eerence (Hypothesis 7) was not

    statistically signicant (p 5 0.10). Whereas neither hypothesis was supported, together the two

    dimensions of emotional labor explained signicant variance in symptoms (DR2 0.04, F(2,

    200) 6.91, p 5 0.001).

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    174 J. SCHAUBROECK AND J. R. JONES

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    13/21

    Table 2. Dimensions of emotional labor predicted by trait aect variables and gender (regression coecients are standa

    Variable name Trait aect hypotheses Variable name G

    Demands toexpress positive

    eerence

    Demands tosuppress negative

    eerence

    Demanexpress p

    eere

    Step 1 Step 1Age 0.09 0.14{ Age 0.1Gender 0.25} 0.12* Education 0.1Education 0.16{ 0.09 Trait negative aect 0.1

    DR2

    0.07; F(3, 207) 5.52} DR2

    0.03; Trait positive aect 0.1F(3, 207) 2.11 Tenure ( job) 0.1Step 2 Job complexity 0.0Trait negative aect 0.03 0.40} Salary 0.1Trait positive aect 0.08 0.19{ DR2 0.10; F(7, 111) 1.81*DR2 0.01; F(2, 205) 0.56 DR2 0.24;

    F(2, 205) 33.84} Step 2Step 3 Gender (1 male; 2 female) 0.2State negative aect 0.13 0.16{ DR2 0.07; F(1, 110) 9.13{State positive aect 0.08 0.02DR2 0.01; F(2, 203) 1.63 DR2 0.02;

    F(2, 203) 2.85

    *p 5 0.10; {p 5 0.05; {p 5 0.01; }p 5 0.001

    Copyright#

    2000JohnWiley&Sons,Ltd.

    J.Organiz.Behav.21,163183(2000)

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    14/21

    Table 3. Eects of emotional labor on physical symptoms (regression coecients are standardized)

    Physical symptoms Physical symptoms Phy

    Step 1Age 0.20}Gender 0.23}Education 0.03Trait negative aect 0.53}DR2 0.43; F(4, 202) 37.27}

    Step 2Demands to express positiveeerence (EPE)

    0.13{

    Demands to suppress negativeeerence (SNE)

    0.12

    DR2 0.04; F(2, 200) 6.91}

    Step 3 Step 3Emotional adaptability (EA) 0.09* Organizational identication (OI) 0.12{DR2 0.01; F(1, 199) 3.10* DR2 0.01; F(1, 198) 4.39{

    Step 4 Step 4 Step 4EPE EA 0.45{ EPE OI 0.15{ SNE OISNE EA 0.17 DR2 0.02; F(1, 197) 6.89{ DR2 0.01; F(1, 197DR2 0.02; F(2, 197) 2.68{

    Note: Where Step 1, 2, or 3 is omitted, the statistics are identical to corresponding steps in previous statement.*p 5 0.10, {p 5 0.05, {p 5 0.01, }p 5 0.001.

    Copyright#

    2000JohnWiley&Sons,Ltd.

    J.Organiz.Behav.21,163183(2000)

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    15/21

    Job involvement and organizational identication were substantially correlated (r 0.67) and

    thus it is not surprising that their respective patterns of relationships with emotional labor were

    essentially isomorphic. Hypothesis 8 predicted that the relationship between emotional labor andphysical health would be stronger among persons reporting little identication with the

    organization or lower job involvement. In omnibus regressions that included all main eects (age,

    gender, education, trait NA, both emotional labor factors, and then either organization

    identication or job involvement), followed by the two product terms, the overall interaction

    block was signicant when organizational identication was the dispositional variable

    (DR2 0.02, F(2, 196) 2.94, p 5 0.05) but not when job involvement was the dispositional

    variable (DR2 0.01, F(2, 196) 1.65, p 5 0.20). Because these variables were signicantly

    correlated, they were examined in separate regression equations. Examined individually, the

    interaction between organizational identication and demands for positive eerence predicting

    physical symptoms was signicant (DR2 0.02, F(1, 199) 6.14, p 5 0.01), as was the case for

    job involvement (DR2 0.01, F(1, 198) 3.98, p 5 0.05). We plotted this interaction by

    inserting cut-point values of / one standard deviation from the mean of each variable into

    the regression equation (Cohen and Cohen, 1983). The plot reveals that whereas positive

    eerence demands were positively related to physical symptoms among persons with low

    organizational identication, there was no relationship between these variables among persons

    reporting high organizational identication (see Figure 2). A largely isomorphic pattern was

    Figure 2. Interaction between perceived demands for positive emotional eerence and organizationalidentication predicting physical symptoms

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    WORKPLACE EMOTIONAL LABOR DIMENSIONS 177

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    16/21

    observed for job involvement, with only the less job-involved persons appearing to suer

    symptoms from perceived demands to express positive eerence. Demands to suppress negative

    emotion did not interact with organizational identication or job involvement in predictinghealth.

    Hypothesis 9 predicted that emotional adaptability would moderate the relationship between

    requirements to express or suppress emotion and physical symptoms. To test this hypothesis we

    rst entered the control variables (trait negative aect, age, gender, and education) and then the

    main eects of the emotional labor indexes. Emotional adaptability was entered at the next step.

    Its incremental main eect on physical symptoms approached statistical signicance, with more

    emotionally adaptive persons reporting fewer symptoms (DR2 0.01, F(1, 199) 3.10,

    p 5 0.08). The block of two interaction terms (the products of emotional adaptability and

    each emotional labor index) was signicant (DR2 0.02, F(2, 197) 2.68, p 5 0.05) and

    therefore this hypothesis was supported (see Table 3). Of the two, only the product term

    involving emotional adaptability and demands to express positive eerence was signicant. The

    plot of this interaction (see Figure 3) indicated that this aspect of emotional labor was strongly

    and positively related to physical symptoms among persons with weaker tendencies in managing

    their emotions. There was virtually no relationship among persons having greater ability to adapt

    their emotions to a given context.

    Figure 3. Interaction between perceived demands for positive emotional eerence and emotionaladaptability predicting physical symptoms

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    178 J. SCHAUBROECK AND J. R. JONES

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    17/21

    Discussion

    There has been very little quantitative empirical research investigating emotional labor in the

    workplace. The present study expanded on the conceptual work and laboratory studies in

    psychology by distinguishing two separate dimensions of emotional labor (demands to express

    positive eerence and requirements to suppress negative eerence) and by testing predictions

    about situational and dispositional determinants and moderators.

    Dispositional correlates

    We observed that perceived demands to express positive eerence were correlated with perceived

    demands to suppress negative emotion, as might be expected, but these variables represented

    separate factors of emotional labor that exhibited discriminant validity. These two dimensions

    were correlated with other variables in dierent ways. Demands for positive eerence was relatedto gender and an occupational `emotional labor' classication (following Hochschild's (1983)

    scheme). Demands for suppression of negative eerence was related to trait aect (PA and NA).

    Thus both situational and dispositional variables appear to be plausible antecedents of perceived

    emotional labor.

    A basic pattern observed within this study was that the personality variables trait NA and trait

    PA predicted perceived demands to suppress negative emotion, whereas they were largely

    uncorrelated with perceived demands to express positive emotion. Conversely, as we summarize

    below, the situational variable of occupational status (independently assessed volume and

    intensity of interpersonal interaction on the job) was correlated with perceived demands to

    express positive emotion but not perceived demands to suppress negative emotion.

    Emotional laborhealth linkages

    The perception of demands to express positive emotion on the job was positively related to

    physical symptoms. The eect of role requirements to express positive emotion on symptoms was

    mainly among individuals who reported little identication with the organization or who were

    less involved with their jobs. The latter nding is consistent with the perspective that emotional

    labor is most unhealthful when one's emotional expressions on the job are not an authentic

    representation of one's personal beliefs.

    One set of factors that may dierentiate whether conformance to an emotional display rule is

    `authentic' is one's trait aect. Whereas these factors did not moderate the relationship between

    perceived emotional labor and physical symptoms, they were related to reports of emotional

    labor in the predicted directions. After accounting for trait negative and positive aect, the

    corresponding state aect variables had little relationship with perceived emotional labor.However, future research might protably examine modal state aect on the job in addition to

    trait aect and aective states around the time the responses are elicited. How one perceives

    emotional labor may be even more strongly inuenced by `job aect' than it is by trait aect.

    Emotional labor seems to be embedded in social and cultural contexts in complex ways.

    `Authenticity' in how people respond to their own emotion work seems to play an important role

    in its eects on health. Hochschild (1983) noted that the psychological eects of emotion work

    among the ight attendants she studied may have been linked to such authenticity perceptions.

    She suggested that when workers do not `own' their emotions and behavior, they suer more

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    WORKPLACE EMOTIONAL LABOR DIMENSIONS 179

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    18/21

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    19/21

    among men. In the present study her occupational coding scheme of emotional labor was

    observed to converge only weakly (albeit to a level of statistical signicance) with perceived

    emotional labor. In the present study we examined the actual experience of emotional labor asperceived by the job incumbent. A signicant relationship between gender and perceived

    demands to express positive emotion persisted even after variables which may account

    substantially for dierences in occupational and social status (i.e., salary, education, age, job

    complexity) were included in the analysis. This may support Hochschild's (1983) view that the

    traditional prevalence of women in jobs that require more emotional labor reects more than just

    a `pre-market' bias whereby women self-select into emotionally-intensive occupations. It is also

    possible that within the same occupation the expectations of customers and co-workers dier

    toward men and women in a way that `genders' the work role in this respect. Finally, women may

    engage in more emotional labor on their own initiative. These explanations are not mutually

    exclusive. The eect of gender was limited to positive eerence demands, and thus this study

    supports Hochschild's (1983) suspicions that women, to a greater extent than men, undertake

    `emotion work that arms, enhances, and celebrates the well-being and status of others' [italics hers;

    p. 165]. However, Hochschild also suggested that `Women are more likely to be presented with

    the task of mastering anger and aggression in the service of ``being nice'' ' (p. 163). In this sample,

    however, we observed only a marginal relationship between gender and demands for suppression

    of negative eerence.

    The role of occupation in

    emotional labor

    Emotional labor was related to various trait and situational variables in complex ways. The fairly

    low magnitude correlations of our indexes with Hochschild's (1983) occupational index is not

    particularly surprising given that such occupational dummy variables are known to be rather

    decient indices of psychological demands (cf. Ganster and Schaubroeck, 1991). A great deal of

    within-occupation variation in emotional labor is to be expected. This study indicates that whatan individual experiences as emotional labor may re ect a complex matrix of determining factors,

    both internal and external to the individual. If this is true, then simpler formulations about the

    role of emotional labor in health may not be protable.

    Study limitations

    The usual caveats about making causal inferences based upon cross-sectional, correlational data

    apply here. For example, relationships between perceived emotional labor and health complaints

    may partially reect a consistency artefact. One might be especially sceptical about correlations

    between a putative stressor that has been shown to be related to trait aect, as prior work has

    shown that relationships between perceived stressors and self-report symptoms are often

    explained by other variables, especially trait negative aect (Brief et al., 1988). However, webelieve the present ndings are quite strongly suggestive of a reliable linkage between emotional

    labor dimensions and physical symptoms for two reasons. First, the tests of their eects on

    symptoms controlled for trait negative aect and demographic risk variables. Second, some of

    our ndings were interaction eects, and these interaction eects followed patterns of theoretical

    expectations that were of such a complexity as to be less plausibly explained by response set biases

    such as priming or consistency. Nevertheless, several of our hypothesis tests involved simple

    linear relationships and therefore common method variance is a possible threat to internal

    validity. In addition, we focused on only a single organization and a limited range of occupations.

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    WORKPLACE EMOTIONAL LABOR DIMENSIONS 181

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    20/21

    We must also caution against generalizing from our study, which predicted somatic symptoms in

    an occupationally diverse service organization, to a broader range of health disorders and to

    particular occupations.

    Conclusions

    The results of this study suggest that it may be necessary for researchers investigating emotional

    labor to distinguish conceptually and empirically between perceptions of required eerence and

    imputed characteristics of the job or occupation. The extent to which individuals perceive that

    they are required to express or suppress certain types of emotional expression may depend as

    much on their emotional predispositions as it does on the objective characteristics of their

    organizational roles. Although the present study did not exhaust the list of all possible individual

    dierence antecedents, the relationships we observed between individual dierences and

    perceived emotional labor were not so strong as to suggest that emotional labor is some form

    of psychological epiphenomenon. It may be that the characteristics of one's role requirements

    actually dier substantially among incumbents of the same job title. Future research may

    protably investigate the extent to which dierences in the expectations held by others determine

    self-perceptions of emotional labor. This would require collecting data from peers, supervisors,

    perhaps other co-workers, and even customers or other types of clients.

    Our study also observed that the extent to which people perceive that they are required to

    express or suppress emotion on the job has some adverse eect on physical health. Moreover, to

    the extent that people identify closely with their roles and/or they have diculty in adjusting their

    emotional expressions to the social context, the eect of perceiving such requirements is more

    unhealthful. Increasingly organizations put pressure on workers to maintain a positive `face' to

    the public and others in the workforce. If workers are to thrive in such service-orientedenvironments they would benet from training and/or selection practices that focus on emotion

    management. As noted by Ashkanasy (1996), emotional displays have import not just to the

    actor himself or herself but also to those individuals who must attempt to gauge and act upon

    what they infer from others' emotional displays. Thus if individuals can be supported in their

    attempts to manage their emotions in healthy ways, it is likely that their emotional displays will

    be more authentic. This may enhance the overall healthfulness of the work environment by

    improving the quality of interpersonal relationships. Future research should therefore explore not

    only the individual consequences of emotional labor, such as health symptoms and satisfaction,

    but how perceived emotional display rules inuence the quality of the interpersonal environment

    by inuencing social perception processes.

    References

    Adelmann PK, Zajonc RB. 1989. Facial eerence and the experience of emotion. Annual Review ofPsychology 40: 249280.

    Ashforth BE, Humphrey RH. 1993. Emotional labor in service roles: the inuence of identity. Academy ofManagement Review 18: 88115.

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    182 J. SCHAUBROECK AND J. R. JONES

  • 7/28/2019 Antecedents of Workplace Emotional Labor

    21/21

    Ashkanasy NM. 1996. Distinguishing felt from displayed emotion in the workplace: implications formanagement research. Paper presented at the annual meeting of the Academy of Management,Cincinnati, USA.

    Brief AP, Burke MJ, George JM, Robinson BS, Webster J. 1988. Should negative aectivity remain anunmeasured variable in the study of job stress? Journal of Applied Psychology 73: 193198.

    Camman C, Fichman M, Jenkins GD Jr, Klesh JR. 1969. Assessing the attitudes and perceptions oforganization members. In Assessing Organizational Change, Goodman PR (ed.); Wiley: New York.

    Caplan RD, Cobb S, French JRP, Harrison R, Pinneau SR. 1975. Job demands and worker health.(NIOSH publication no. 75-160). Washington, DC: Department of Health, Education, and Welfare.

    Cohen J, Cohen P. 1983. Applied Multiple Regression Analysis in the Behavioral Sciences. Erlbaum:Hillsdale, NJ.

    Cohen S, Herbert TB. 1996. Health psychology: psychological factors and physical disease from theperspective of human psychoneuroimmunology. Annual Review of Psychology 47: 113142.

    Ekman P. 1972. Universals and cultural dierences in facial expressions of emotion. In NebraskaSymposium on Motivation, Cole JK (ed.); University of Nebraska Press: Lincoln, NE; 207283.

    Friedman HS. 1989. The role of emotional expression in coronary heart disease. In In Search of Coronary-prone Behavior: Beyond Type A, Siegman AW, Dembroski TM (eds); Erlbaum: Hillsdale, NJ; 149168.

    Ganster DC, Schaubroeck J. 1991. Work stress and employee health.Journal of Management 17

    : 235271.Gordon SL. 1989. Institutional and impulsive orientations in selectively appropriating emotions to self. In

    The Sociology of Emotions: Original Essays and Research Papers, Franks DD, McCarthy ED (eds); JAIPress: Greenwich, CT; 115135.

    Hochschild AR. 1979. Emotion work, feeling rules, and social structure. American Journal of Sociology 85:551575.

    Hochschild AR. 1983. The Managed Heart: Commercialization of Human Feeling. University of CaliforniaPress: Berkeley, CA.

    House JS. 1980. Occupational Stress and the Physical and Mental Health of Factory Workers. Institute forSocial Research: Ann Arbor, MI.

    Izard CE. 1971. The Face of Emotion. Appleton-Century-Crofts: New York.King LA, Emmons RA. 1990. Conict over emotional expression: psychological and physical correlates.

    Journal of Personality and Social Psychology 58: 864877.Laird J, Crosby M. 1974. Individual dierences in self-attribution of emotion. In Thought and Feeling:

    Cognitive Alteration of Feeling States, London H, Nisbett RE (eds); Aldine: Chicago; 4459.

    Morris JA, Feldman DC. 1996. The dimensions, antecedents, and consequences of emotional labor.Academy of Management Review 21: 9861010.

    Morris JA, Feldman DC. 1997. Managing emotions in the workplace. Journal of Managerial Issues IX:257274.

    Notarious CI, Levenson RW. 1979. Expressive tendencies and physiological response to stress. Journal ofPersonality and Social Psychology 37: 12041210.

    Notarious CI, Wemple C, Ingraham LJ, Burns TJ, Kollar E. 1982. Multichannel responses to aninterpersonal stressor: interrelationships among facial display, heart rate, self-report of emotion, andthreat appraisal. Journal of Personality and Social Psychology 43: 400408.

    Rafaeli A, Sutton RI. 1987. Expression of emotion as part of the work role. Academy of ManagementReview 12: 2337.

    Sutton RJ. 1991. Maintaining norms about expressed emotions: the case of bill collectors. AdministrativeScience Quarterly 36: 245268.

    Sutton RI, Rafaeli A. 1988. Untangling the relationship between displayed emotions and organizationalsales: the case of convenience stores. Academy of Management Journal 31: 461487.

    Watson D, Clark LA. 1984. Negative aectivity: the disposition to experience negative emotional states.Psychological Bulletin 96: 465490.

    Watson D, Clark LA, Tellegen A. 1988. Development and validation of brief measures of positive andnegative aect: the PANAS scales. Journal of Personality and Social Psychology 54: 10631070.

    Wharton AS. 1993. The aective consequences of service work: managing emotions on the job. Work andOccupations 20: 205232.

    Wharton AS, Erickson RJ. 1993. Managing emotions on the job and at home: understanding theconsequences of multiple emotional roles. Academy of Management Review 18: 457486.

    Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 163183 (2000)

    WORKPLACE EMOTIONAL LABOR DIMENSIONS 183