International Journal of Hospitality Management 31 (2012) 1067–1082

Contents lists available at SciVerse ScienceDirect

International Journal of Hospitality Management

journal homepage: www.elsevier.com/locate/ijhosman

The managed heart: The structural analysis of the stressor–strain relationship

and customer orientation among emotional labor workers in Korean hotels
Gyehee Lee a,1 , Taegoo (Terry) Kim a,∗ , Seok Ho Shin b,2 , Ick Keun Oh c,3
a
Department of Tourism Sciences, College of Hotel and Tourism, Kyung Hee University, 1 Hoegi-dong, Dongdaemun-gu, Seoul 130-701, Republic of Korea
b
Department of Food Service Management, College of Natural Resources, Young Nam University, 214-1 Dae-dong, Gyeongsan-si, Gyeongsangbuk-do 712-749, Republic of Korea
c
Department of Tourism Studies, College of Business, Keimyung University, 2800 Dalgubeoldaero, Dalseo-gu, Daegu 704-701, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Keywords: This study examined the structural relationships among three different dimensions of workplace stress-
Workplace stressor ors (customer-related stressor, CRS; work environment-related stressor, WERS; job-related stressor, JRS),
Emotional labor negative affectivity (NA), emotional exhaustion (EE), and the negative effect of that strain on customer
Negative affectivity
orientation (CO) in the context of the emotional labor (EL) of frontline employees in the hotel industry.
Emotional exhaustion
Data were collected from self-administrated questionnaires distributed among frontline employees in
Customer orientation
Conservation of resources theory room and F&B divisions in Korean deluxe tourist hotels, where EL is intense. The results of the structural
Human resource management equation analysis indicated a positive association between all three workplace stressors and NA and
Marketing strategy between NA and EE. There was also a relationship in the opposite direction with EE and CO, as hypoth-
Hotel frontline employees esized. The moderating effect of organizational level on the workplace stressors–NA relationships was
Korea also confirmed. In addition, in an alternative model, we found that NA partially mediates the relationship
between JRS and EE; whereas, NA fully mediates the relationships between CRS/WERS and EE. Practical
implications are discussed in detail and limitations of the study and future research directions are also
suggested.
© 2012 Elsevier Ltd. All rights reserved.

1. Introduction management? If so, how are frontline employees’ emotions man-

The biggest issue for human resource management (HRM) of
organizations is to socialize their frontline employees by shap-
ing and directing their behavior. Most training programs in the
hotel industry are geared toward achieving this goal. Pugh’s (2001)
research revealed that frontline employees can transmit their emo-
tions to customers in a service encounter. Thus, positive emotional
displays by frontline employees lead to positive emotive experi-
ences, resulting in higher customer evaluations of service quality.
In the hotel industry, service providers’ emotional displays are
pivotal to successfully creating and delivering a delightful ser-
vice experience. This is directly linked to a service organization’s
bottom-line, as suggested by the service-profit chain that links
employee attitudes, customer satisfaction, and organizational prof-
its (Babakus et al., 2009). Given that the “performance” of frontline
employees is a managerial object, are their emotions also objects of
∗ Corresponding author. Tel.: +82 2 961 2321; fax: +82 2 964 2537.
E-mail addresses: ghlee@khu.ac.kr (G. Lee), tgkim@khu.ac.kr (T. Kim),
shs77@hanmail.net (S.H. Shin), ikok@kmu.ac.kr (I.K. Oh).
1
Tel.: +82 2 961 0863.
2
Tel.: +82 53 591 6077.
3
Tel.: +82 53 580 6403.
aged? This is a particularly critical concern for the hospitality
industry due to the inseparability of the end product from those
who produce it.
Hochschild (1983) coined the term “emotional labor (EL),” refer-
ring to the expression of emotions and creation of feelings as an
expected part of work roles. EL requires “face-to-face or voice-
to-voice contact with the public, and the workers are required
to produce an emotional state in another person and allow the
employer to exercise a degree of control over the emotional activ-
ities of frontline employees” (Hochschild, 1983, p. 147). In the
fiercely competitive industry of full service luxury hotels, the EL
expected of frontline employees is intense due to the high expec-
tations of customers and diminishing differences in facilities and
amenities among numerous competitors. Therefore, the quality
of service transactions in such hotels is often determined by the
frontline employees’ personal service (Pugh, 2001). This requires a
higher standard of emotional display rules for frontline employees.
Hotel management should take a serious look at EL issues from
two perspectives that are traditionally distinguished: HRM and
marketing. This integrated perspective provides hotel managers
with critical insight into how to create customer equity through
effective management of their human capital. Given that front-
line employees are the most critical element of a hotel marketing
strategy, examining the workplace stressor–strain relationship and

0278-4319/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhm.2012.01.003
1068 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
its consequence for hotel employees should be a major concern
for scholarly researchers and marketing practitioners alike (Singh,
2000). Frontline employees are direct participants in implementing
the marketing concept (Brown et al., 2002), and their attitudes and
behaviors toward customers determine customers’ perceptions of
service quality and satisfaction, which in turn affects organiza-
tional performance (Rust et al., 1996). A vast array of empirical
evidence exists on the direct influence of emotional exhaustion
(EE) on outcomes such as lower job performance (e.g., Cropanzano
et al., 2003; Karatepe and Aleshinloye, 2009), lower job satisfaction
(e.g., Karatepe, 2006; Karatepe et al., 2009; Mulki et al., 2006), and
higher turnover intentions (e.g., Babakus et al., 2008; Cropanzano
et al., 2003; Karatepe and Aleshinloye, 2009; Karatepe et al., 2009)
in various fields. However, the dysfunctional influence of EE on
customer orientation (CO) has been left largely untouched. The
paucity of research is due to the fact that EL issues have been
treated mainly within the areas of HRM and organizational behav-
ior (OB), but not critically from a marketing perspective. A recent
exception to this is a study by Julian (2008) that explored criti-
cal managerial/marketing issues related to EL using case studies.
It was concluded that CO is a key outcome variable and deserves
more attention from both marketing researchers and practitioners
alike. Therefore, this study attempts to close the research gap by
expanding a stressor–strain model postulating CO as a direct out-
come variable. Integrating HRM and marketing in this study allows
us to go beyond the negative effects of EL on personal level and
examine the structural relationship among workplace stressors,
strain outcome, and the effectiveness of hotels in terms of their
marketing performance.
Another meaningful contribution of this study is the examina-
tion of the effects of an entire array of workplace stressor factors
on employees’ strain which can take various progressive forms,
including emotional, psychological, physical, social, and behavioral
outcomes (for the entire facets of strain, see Beehr and Newman,
1978). In numerous studies, diverse stressors have been identified.
However, most of these deal with the stress factors sporadically. A
multi-disciplinary literature review in such areas as organizational
studies, marketing, psychology, and sociology converged into three
main domains of workplace stressors that evoke NA in employees
and lead to EE in a customer encounter: customer-related stressor
(CRS) (Ashforth and Humphrey, 1993; Cordes and Dougherty, 1993;
Dormann and Zapf, 2004; Fullerton and Punj, 2004; Grandey et al.,
2007; Karatepe et al., 2010), work environment-related stressor
(WERS) (Ashforth and Humphrey, 1993; Pugh, 2001; Rafaeli and
Sutton, 1987; Sutton and Rafaeli, 1988), and job-related stressor
(JRS) (Chen and Spector, 1991; Jamal, 2004; Morris and Feldman,
1996; Ross, 1993). These studies, however, tested their effects sep-
arately. We simultaneously tested the entire array of workplace
stressor factors in a single model. Accurate information regarding
the comparative effects of such factors will allow hotel managers
to effectively cope with workplace stressors.
Another important research question we explored is the mod-
erating effect of organizational level (OL) on the workplace
stressors–NA relationship. The main purpose of our model spec-
ification regarding the moderating role of OL is to provide
theoretical and practical implications. The stressor–NA relation-
ship and its outcome must be different across rank and file level
(RFL) employees and over supervisory level (OSL) employees. Cop-
ing mechanisms and resources used by RFL employees and OSL
employees of hotels may not be the same. There seems to be
a justification for different treatment and programs designed to
alleviate the EL and its negative consequence for the RFL employ-
ees and OSL employees separately. This approach is of particular
use for managers of the hotel industry. They need to under-
stand how and why a certain stressor intensifies the EL of their
employees in order to develop coping strategies and training
programs tailored to employees’ needs at different organizational
ranks.
The principal objective of this study is to test a theoretical
model purported to analyze the structural relationships among
key constructs related to EL, including NA and its antecedents (i.e.,
workplace stressors: CRS, WERS, and JRS), EE, and its dysfunctional
effect on CO. The specific objectives of this study are four-fold:
(1) to identify and categorize workplace stressors that precede the
NA of hotel frontline employees and examine the relative effect
of each factor directly related to NA; (2) to examine the structural
relationships among NA, EE, and CO; (3) to investigate the moderat-
ing role of OL, RFL vs. OSL, on the relationship between workplace
stressors and employees’ NA; and (4) to examine the mediating
role of employees’ NA between workplace stressors and EE in an
alternative model.
The findings of this study can serve as a theoretical basis for
EL research in hotels with respect to the workplace stressor–strain
relationship and its consequence from both an HRM and marketing
perspectives.
2. Conceptual model and hypotheses
The research model of this study postulated a structural analysis
among EL workers’ workplace stressors, NA, EE, and CO in hotels,
as depicted in Fig. 1. In the marketing literature, stress factors have
been sporadically suggested but empirical tests that examine those
factors concomitantly have been rare. As mentioned previously, a
multi-disciplinary literature review in such areas as organizational
studies, marketing, psychology, and sociology, converged into three
main domains of workplace stressors that may affect NA in a cus-
tomer encounter, namely CRS, WERS and JRS. The three workplace
stressors were thus incorporated in our research model.
Studies on factors that cause NA are mainly in the fields of psy-
chology and OB. A review of the literature found that these studies
converge into two main domains of stressor precursors, overall
organizational factors and personal traits. Given that employees’
EE is believed to occur most frequently in service encounters
(Babakus et al., 2008; Cropanzano et al., 2003; Karatepe and
Aleshinloye, 2009; Phillips et al., 2007), identifying factors that
influence employees’ EE and, ultimately, CO is a potentially fruit-
ful venue for HRM and marketing research integration. This study
adopted CO as a direct outcome variable of EE. As suggested by
Narver and Slater (1990), in this study CO was measured as how
willing hotel frontline employees are to meet customers’ needs
instead of their own. This exerts a direct effect on service quality.
This study tests hypotheses regarding the moderating effect of
OL on the relationship between workplace stressors and NA. Most
previous empirical studies did not take individual characteristics,
such as an employee’s OL, responsibilities, years of employment,
or functional areas of work into consideration. However, as an
individual’s career develops he or she tends to accumulate more
formal and informal resources to cope with workplace stressors and
reduce consequential strains, such as burnout (Beehr et al., 1990).
Thus, OL appears to be a critical factor to examine the moderating
effect on the workplace stressor–NA relationship.
2.1. Workplace stressors and negative affectivity
Affectivity refers to a general tendency to experience a partic-
ular mood or react to objects in a particular way or with certain
emotions (Lazarus, 1993). Although there are similar terms like
emotions and moods, affectivity is considered a general cate-
gory of mental feeling processes, including emotions, moods, and
attitudes (Bagozzi et al., 1999). The two theoretically and empir-
ically distinctive dispositional personality variables are positive
 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
Organizational Level
(RFL vs. OSL)
H4a
Customer-
related
Stressor
H1a (+)
H4b
Workplace Stressors
Work
Environme
nt-related
Stressor H1b (+)
H4c
Job-related H1c (+)
Stressor
Human Resource Management Perspective
Note: RFL, rank and file level; OSL, over supervisor level.
affectivity (PA) and NA (Agho et al., 1992). NA is the tendency to
experience negative emotional states such as sadness, anxiety, and
hostility. The definition NA adopted in this study is a “dispositional
dimension which reflects a tendency to view things negatively”
(Watson and Clark, 1984, p. 465). Individuals high on PA are more
likely to experience positive affective states over time and across
a variety of situations. They tend to have an overall sense of well-
being, to see themselves as active, self-efficacious, pleasurably and
effectively engaged, and to experience positive emotional states
(George, 1989). On the other hand, high NA individuals generally
tend to perceive events and individuals in a more negative man-
ner and are anxious and nervous (Iverson et al., 1998). Therefore,
individuals who are high on NA tend to be easily distressed, agi-
tated, upset, pessimistic and dissatisfied. They have a negative view
of themselves and the world and are more introspective. These
individuals tend to dwell on their past failures and shortcomings
(Watson and Clark, 1984).
In the last three decades, NA has received considerable research
attention in various fields, including HRM, OB, marketing, psychol-
ogy, sociology, and nursing (Agho et al., 1992; Anderson et al.,
2002; Cordes and Dougherty, 1993; Dormann and Zapf, 2004;
Fullerton and Punj, 2004; Iverson et al., 1998; Jamal, 2004; Karatepe
and Uludag, 2008; Karatepe and Aleshinloye, 2009; Ross, 1993;
Watson and Clark, 1984; Young and Corsun, 2009). Morris and
Feldman (1996) highlighted that NA frequently occurs during ser-
vice encounters for service providers. NA entails direct and indirect
negative emotional/behavioral outcomes, such as EE, high turnover,
and low job satisfaction. A few theorists and empirical study results
provide a solid guideline to develop the hypothesis regarding the
role of NA as mediator (Bradburn, 1969; Diener and Emmons,
1984; Watson and Clark, 1984). As those theorists suggest, people
with more NA have tendency to experience more severe discom-
fort in stressful situations. This NA disposition may leads to high
turnover and low job satisfaction (Agho et al., 1992, 1993; Brief
et al., 1988; Levin and Stokes, 1989; Staw and Ross, 1985; Staw
1069
H2 (+) H3 (−)
Negative Emotional Customer
Affectivity Exhaustion Orientation
Marketing Perspective
Fig. 1. Hypothesized research model.
et al., 1986; Watson et al., 1989). Liu et al. (2004) empirically found
that individuals with high NA have a tendency to feel more neg-
atively under workplace stressors and this dispositional tendency
impede them to regulate their emotional experiences in the ser-
vice encounters. Whereas earlier researchers have looked into the
bivariate relationship between situational variables and turnover,
and job satisfaction, later researchers began their investigations on
stressors and their effects on personality variables (NA and PA), and
thus NA as a personality dimension has been widely used in strain-
related research (Houkes et al., 2001). For instance, in explaining
the direct effects of workplace stressors on NA, Van De Vliert and
Van Yperen (1996) found that employees can experience more NA
when workplace stressors increase. Spector et al. (2000), based on
the perception mechanism, stated that high NA individuals tend
to perceive their jobs as more stressful and experience high lev-
els of strains. Thoresen et al. (2003) reported a strong correlation
between NA and EE. Considering the existing literature, it is gener-
ally believed that NA is frequently evoked in EL situations and is a
significant antecedent to strains such as emotional dissonance and
EE, and job dissatisfaction (Karatepe and Aleshinloye, 2009).
In more recent years, the role of NA as a mediator or a moderator
has been tested in complex stressor–strain models. For instance, in
Zapt’s (2002) and Dallimore et al.’s (2007) research, NA also has
been used as a mediator in stressor-strain relationship. Zapf (2002)
delineated that NA exacerbated by EL has a direct positive effect
on EE and depersonalization. NA is evoked by stress factors, espe-
cially when employees are involved more with surface acting and it
tends to lead to EE and job dissatisfaction, and other negative out-
comes (Brotheridge and Grandey, 2002; Brotheridge and Lee, 2003;
Dallimore et al., 2007). These previous studies readily provide a
workable guideline for our study hypothesis regarding the medita-
tional role of NA in the stressor–strain relationship. We, therefore,
hypothesize that frontline employees in the hotel industry with
high NA are susceptible to stressors and this dispositional tendency
may lead to higher EE (Karatepe and Aleshinloye, 2009).
1070 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
As previously mentioned, this study adopted three dimensions
of frontline employee workplace stressors, namely CRS, WERS, and
JRS, which may lead to NA. CRS includes disproportionate customer
demands, verbal aggression, and negative attitude or behavior
toward service providers (Dormann and Zapf, 2004; Fullerton and
Punj, 2004; Karatepe et al., 2010). These aggressive behaviors and
attitudes are identified as a serious issue in the workplace. Dis-
proportionate customer demands or bad attitudes cause negative
appraisal and NA for employees (Fullerton and Punj, 2004). Front-
line employees in service industries frequently encounter negative
customer reactions and verbal aggression, which can make them
even more prone to NA and eventually EE (Cordes and Dougherty,
1993; Dormann and Zapf, 2004; Grandey et al., 2007; Karatepe et al.,
2010). WERS consists of problematic situations and environments,
such as not having enough rest areas, high temperatures, bad illu-
mination, and crowding in the work place (Ashforth and Humphrey,
1993; Kim and Suh, 2005; Sutton and Rafaeli, 1988). Bitner (1992)
pointed out the insufficiency of empirical research regarding the
influence of atmospherics on employees. The physical environment
influences not only customers’ emotive, cognitive, and psycholog-
ical responses but employees’ as well. This can lead to avoidance
behaviors in service encounters via NA. JRS, such as high workload,
unfavorable work hours, and noncooperation/nonsupport of supe-
riors and colleagues has been documented to contribute to frontline
employees’ NA (Anderson et al., 2002; Morris and Feldman, 1996).
Employees who face JRS, such as a high workload, often suffer from
EE and experience a high need of recovery (Zapf et al., 2001) because
they have to invest extra effort and regulate their emotions in order
to meet job demands (Anderson et al., 2002; Rodriguez et al., 2001).
Similarly, irregular work schedules often evoke frontline employ-
ees’ NA, and consequently contribute to EE. Previous research
shows that working a nonstandard schedule is related to employ-
ees’ negative attitudes toward their jobs and organization, which
leads to low organizational commitment, low job satisfaction, and
high turnover (Furnham and Hughes, 1999). For instance, Jamal’s
(2004) empirical study based on full-time Canadian employees at
a variety of organizations showed that employees on nonstandard
schedules experience higher levels of NA, turnover, and EE than
those working a 9–5 Monday through Friday schedule. Ross (1993)
pointed out that a supportive atmosphere eases the interpersonal
pressure on restaurant service employees when they interact with
customers and leads to reduced NA. In light of the preceding dis-
cussion, we advance the following hypotheses:
Hypothesis 1a. CRS has a positive effect on NA among hotel front-
line employees.
Hypothesis 1b. WERS has a positive effect on NA among hotel
frontline employees.
Hypothesis 1c. JRS has a positive effect on NA among hotel front-
line employees.
2.2. Negative affectivity and emotional exhaustion
Conceptually, strain has been defined as “a set of psycho-
logical and physiological responses to stressors, including long
term [chronic] problems with physical and mental well-being”
(Stone-Romero, 1994, p. 160) and includes the following three
broad categories of responses: physical, psychological, and behav-
ior (Spector et al., 2000). EE is an example of a psychological strain
(Beehr et al., 1990; Gaines and Jermier, 1983) and it is defined as
“the feeling of being emotionally overextended and exhausted by
one’s work” (Maslach and Jackson, 1981, p. 101). It often occurs
when employees perform “people-work” (Cordes and Dougherty,
1993). EE signifies the experience of being strained and depleted of
one’s emotional and physical resources (Karatepe and Aleshinloye,
2009; Maslach, 1982; Zapf, 2002). A psychological syndrome con-
nected to chronic exposure to stressful interpersonal encounters in
work contexts, it is the most prominent facet of job burnout.
NA is widely used in strain-related research and has been
linked with EE (e.g., Houkes et al., 2001; Judge et al., 2009;
Karatepe and Aleshinloye, 2009). Iverson et al. (1998) hypothesized
a causal model of affectivity, organizational stressors, burnout,
and absenteeism and showed that NA results in high levels of
EE, depersonalization, and low personal accomplishment. Using
meta-analysis, Thoresen et al. (2003) reported an estimated mean
population correlation of 0.54 between NA and EE. In hospital-
ity research, Karatepe and Aleshinloye (2009) reported that NA is
positively related to EE among hotel employees. Thus, in the cur-
rent study hotel frontline employees with high NA are believed
to be susceptible to high level of EE. In light of the above dis-
cussion and prior empirical findings, the following hypothesis is
proposed:
Hypothesis 2. NA has a positive effect on EE among hotel frontline
employees.
2.3. Emotional exhaustion and customer orientation
The marketing literature has traditionally explored the
antecedents of customer-oriented behaviors (Hartline et al., 2000;
Hoffman and Ingram, 1991). CO is a fundamental concept to the the-
ory and practice of marketing management. It is well established
that customer-oriented firms outperform competitors and provide
greater customer satisfaction (Hoffman and Ingram, 1991; Julian,
2008; Kim and Suh, 2005; Kim and Kwon, 2010). Recently, Brown
et al. (2002, p. 111) conceptualized CO at an individual level as “an
employee’s tendency or predisposition to meet customer needs on
the job.” In addition, the marketing literature has conceptualized
CO as beliefs and surface personality traits rather than behaviors
(Brown et al., 2002; Kennedy et al., 2002; Licata et al., 2003). Limited
but growing empirical evidence indicates that CO directly affects
frontline employees’ job performance and also countervails against
EE (Babakus et al., 2009).
The postulated relationship between EE and CO is based on
the fact that EE experienced during customer interactions greatly
influences employees. It can cause maladjustment to a job, apathy
toward coworkers and customers, self-alienation from work, reluc-
tance to engage in helpful and participative behavior, and a lack of
bonds with others (Julian, 2008; Kim and Kwon, 2010). These neg-
ative employee reactions ultimately deteriorate CO (Julian, 2008;
Kim and Suh, 2005; Loseke and Cahill, 1986; Thoits, 1986). More-
over, EE causes dysfunctional psychological (e.g., little concern for
others or decreased altruism) and physical (e.g., intense sense of
weariness and depression) effects and increases the possibility of
negative behavior (Kim and Kwon, 2010; King and Emmons, 1990).
Individuals experiencing EE try to maintain a psychological sense of
distance from others as a coping strategy (Maslach, 1982) and often
experience limited patience during face-to-face or voice-to-voice
interactions with customers. More recently, Kim and Kwon (2010)
empirically tested the structural relationships among Korean hotel
frontline employees’ emotional dissonance, EE, and CO. The results
showed that employees’ EE significantly decrease their CO. There-
fore, when frontline employees experience EE, customer oriented
attitudes and behaviors, such as an interest in customers’ needs,
tend to decrease. In summary, frontline employees’ EE during inter-
actions with customers has an adverse effect on their CO. Thus, we
proposed the following hypothesis:
Hypothesis 3. EE has a negative effect on CO among hotel frontline
employees.
 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
2.4. The moderating role of organizational level
OL has been treated as one of the controlling variables, along
with other demographic variables such as gender, age, organiza-
tional tenure, educational level, and marital status to partial out
the confounding effects in the stressor–strain relationship (e.g.,
Karatepe and Aleshinloye, 2009; Karatepe et al., 2009). A thorough
review of the existing literature reveals that since the early stages
of EL research, scholars have examined the moderating effect of
OL on the stressor–strain relationship (e.g., Brymer et al., 1991;
Fisher and Gitelson, 1983; Hamner and Tosi, 1974; Zohar, 1994). For
instance, Hamner and Tosi (1974) found that lower level employ-
ees experience more role conflict, presumably because they lack
the power to resolve it. Using meta-analysis, Fisher and Gitelson
(1983) showed a stronger negative association between role ambi-
guity and job satisfaction in lower level jobs than in management
jobs. One explanation for this moderating effect is that employ-
ees in higher positions tend to better manage their role ambiguity
since they have more resources (e.g., power, autonomy, and deci-
sion latitude) to cope with unclear situations. Brymer et al. (1991)
also found that workplace stressors, such as high workload and role
conflict, cause more dysfunction for lower level employees because
they have less formal authority and lower relative status. However,
other researchers argued that the adverse effect of role conflict may
be more salient for managers since they have higher expectations
in terms of their work-role involvement (Zohar, 1994).
Among more recent studies, Karatepe’s (2011) study of hotel
frontline employees in Turkey tested the moderating role of
perceived organizational support and job autonomy as a proxy
moderator on EE. He concluded that the two moderator variables
act as organizational resources: job resources and extrinsic moti-
vation. Therefore, the effects of such variables buffer their negative
effect on EE. Karatepe (2011) argued that OL may play an important
role in determining the structural relationship among exogenous
and outcome variables. In other words, the structural relationship
between stressors and strains may vary depending on the employ-
ees’ OL. B.P. Kim et al. (2009) also confirmed the moderating effect of
OL between stressors and strains. This was also confirmed by other
previous research (e.g., Babin and Boles, 1998; Fisher and Gitelson,
1983). As Kim and his colleagues pointed out, our understanding of
its role is still very limited. Therefore, this study explored the mod-
erating effect of OL in the more diverse context of the hospitality
industry. Thus, it is a logical extension of the previous findings to
hypothesize that as hotel employees move up the organizational
hierarchy to upper managerial levels, they tend to receive more
job autonomy and access to more resources that buffer negative
outcomes (Karatepe, 2011).
From theoretical perspective, the conservation of resources
(COR) theory suggested by Hobfoll (1989) justifies our hypoth-
esis regarding the moderating role of OL. The COR theory is a
highly endorsed theoretical framework for EL (Karatepe et al.,
2009, 2010), which states that job resources alleviate employ-
ees’ stressors and strains. According to the COR theory, individuals
accumulate resources they can apply to accommodate, withstand,
or overcome the threat of losing resources. These include per-
sonal (i.e., self esteem), material (i.e., money), and conditional (i.e.,
status and social support) resources or energies (Hobfoll, 1989),
which provide means efficacy, social support, and perceived con-
trol over stressful conditions (Chen et al., 2009). Among several
stress related theories, the COR theory, bridging the gap between
environmental (stress as environmentally produced) and cognitive
(stress as individual cognition) viewpoints, may explain how vari-
ous forms of resources can help reduce EL stress. Therefore, we can
expand this logic to assume that as an individual’s career develops,
employees tend to accumulate more resources and better access
to them in order to cope with workplace stressors and reduce
1071
consequential strains which can take various progressive forms,
including emotional, psychological, physical, social, and behav-
ioral outcomes (Beehr and Newman, 1978). Such resources may
include organizational and/or peer supports, respect from their
subordinates and guests, family support, job autonomy and sta-
tus, training, rewards, guest handling skills accumulated from work
experiences, and psycho-emotional resilience developed from their
own work-life experiences (Cohen and Willis, 1985; Cohen and
Edwards, 1989; Hobfoll, 1989). The COR theory also explains, in
part, why RFL employees tend to have more negative experiences
than manager level employees. According to the COR theory, “..
resource loss is disproportionately more salient than resource gain”
(Hobfoll, 2001, p. 343). Once frontline employees recognize that
stressors threaten their resources or that resources are lost in the
process of juggling these stressors, they tend to develop NA easily
and eventually become emotionally exhausted.
Therefore, in line with the COR theory and related empirical
findings, we hypothesized that OL is an important moderator. That
is, managerial employees in a hotel are less vulnerable to stress fac-
tors and their NA is less severely affected as compared with lower
level employees. The above discussion and literature provides jus-
tification for the following hypothesis:
Hypothesis 4a. OL exhibits a moderating effect on the relationship
between CRS and NA among hotel frontline employees; specifically,
higher CRS is associated with greater NA for RFL employees than
for OSL employees.
Hypothesis 4b. OL exhibits a moderating effect on the relation-
ship between WERS and NA among hotel frontline employees;
specifically, higher WERS is associated with greater NA for RFL
employees than for OSL employees.
Hypothesis 4c. OL exhibits a moderating effect on the relationship
between JRS and NA among hotel frontline employees; specifically,
higher JRS is associated with greater NA for RFL employees than for
OSL employees.
3. Method
3.1. Measurement
All measurements were drawn from previous research and
aligned with the conceptual aspects of each construct in this
study setting. The survey instrument was originally prepared in
English and then translated into Korean via back-translated method
(Parameswaran and Yaprak, 1987). Responses to the scale items for
the three workplace stressors (i.e., CRS, WERS, and JRS), NA, EE, and
CO were elicited using a 5-point Likert-type scale ranging from 1
(strongly disagree) to 5 (strongly agree).
Since there are no previous comprehensive examples in the
hotel literature regarding measurement items to operationalize
workplace stressors, we conducted in-depth interviews with hotel
managers. Based on these interviews, we modified existing items
in the literature of other fields and identified new ones relevant to
hotel workplaces. A total of 10, 1–2 h interviews were conducted
with managers with more than 10 years of experience in the hotel
industry. After the initial item pool was finalized, a review was
conducted with hospitality professors and hotel human resource
managers. The final pool contained the 14 most salient items that
may cause frontline employees’ NA. To validate those items, a pilot
test with 70 hotel frontline employees working in the F&B, front
desk, and housekeeping departments were conducted prior to the
actual survey. An exploratory factor analysis (EFA) was applied
to the data to identify dimensionality and examine item load-
ings. As a result of the analysis, three distinctive domains, CRS,
WERS, and JRS, emerged and the 14 items were loaded on one of
1072 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082

the three factors, with loadings above 0.60. The internal consis- Table 1
Profiles of the respondents (n = 399).
tency of all three domains was measured by Chronbach’s alpha and
ranged from 0.762 to 0.865. Four scale items from Dormann and Characteristics Frequency (n) Percentage (%)
Zapf (2004) were used to measure CRS. CRS was conceptualized Gender
as “negative customer-related factors, such as negative customer Male 187 46.9
behaviors, needs, and attitudes that intentionally or accidentally Female 212 53.1
have a negative influence during service encounters.” Based on Age (years old)
Bitner’s (1992) servicescape, which argued that a negative work Less than 30 218 54.6
environment could affect both customers and workers alike, WERS 30–39 142 35.6
40–49 35 8.8
includes long work hours and work conditions in service encoun-
50 or older 4 1.0
ters. This was operationalized using a four-item scale. Three items
were modified from Ashforth and Humphrey (1993), Kim and Suh Education level
High school 33 8.3
(2005), and Rafaeli and Sutton (1987). The final item was devel-
2 year collage 225 56.4
oped in this study. JRS was operationalized as “negative job-related University 119 29.8
factors that cause stress, such as work overload, role ambiguity, Graduate school 22 5.5
insufficient work-related support, and the absence of discord con- Department
trol as an effective tension-handling mechanism.” It was measured Rooms 87 21.8
using a six-item scale. Four items were modified from Anderson Food and beverage 289 72.4
et al. (2002) and Rodriguez et al. (2001). The other two items were Others 23 5.8
developed in this study. Tenure in the current hotel (years)
NA was operationally defined in this study as “a hotel front- Less than 1 110 27.6
line employee’s disposition to experience discomfort across time 1 to less than 3 81 20.3
3 to less than 5 41 10.3
and situations” as suggested by Watson and Clark (1984). Three
5 to less than 10 83 20.8
items adopted from Watson and Clark (1984) were used to mea- 10 to less than 15 56 14.0
sure NA. There are reasons why we used only a part of Watson and 15 or more 28 7.0
Clark’s (1984) NA personality scale. First, in our initial pilot-test, Organizational level
some NA scale items did not have reasonable communality of below Rank and file level 326 81.7
0.40 and some were double loaded across factors (e.g., I suffer from Supervisor level 54 13.5
nervousness, I sometimes feel ‘just miserable’ for no reasons: low Over manager level 19 4.8
communality below 0.40; I often find myself worrying about some- Locations of the hotel
thing and, I often lose sleep over much worries: double loaded). In Seoul 187 46.9
addition, during our interviews, several NA items repeatedly men- Daegu 15 3.8
Gyeongju 149 37.3
tioned; the three items used in this study were the most salient Jeju 25 6.3
ones. Some previous studies used only those three items in mea- Others 23 5.8
suring NA. For instance, Karatepe and Aleshinloye (2009) used a
part of the 11-item scale of NA in their hotel employee EL research.
This succinct sub-scale of the original NA scale did not fail to capture Table 1 summarizes the demographic information of the respon-
the core concept of NA in their studies. Thus, a decision was made dents. Of the total 399 respondents, over one-half (53.1%) were
to operate the NA variable with three-item sub-scale of Watson and female and 46.9% were male. In terms of age, 54.6% of the respon-
Clark’s (1984) NA scale. dents were less than 30 years old, 35.6% were 30–39 years old, 8.8%
As an example of a psychological strain, EE was defined as were 40–49 years old, and 1% were 50 years or older. The majority
“an exhausted state of the emotional resources which occurs dur- of the respondents (91.7%) were highly educated, holding at least a
ing interactions with customers.” It was measured by a four-item two year college degree; 5.5% had some graduate education. Of the
scale adapted from Maslach and Jackson’s (1981) Maslach Burnout sample, 20.3% had 1–3 years experience at their current hotel and
Inventory (MBI). 20.8% had 5–10 years experience. Having almost one-half (45.1%) of
Finally, CO, the final dependent variable in the proposed model employees experienced in the current hotel for 3–15 years provided
for this study, was defined as “a hotel frontline employee’s predis- great feedback value on the survey questions. In terms of organi-
position or behavior to meet customer needs on the job.” It was zational hierarchy, the ratio of the sample of RFL, supervisor, and
measured using a five-item scale drawn from Narver and Slater over manager levels was 81.7:13.5:4.8. They were either RFL (326,
(1990). 81.7%) or OSL (73, 18.3%). Overall, based on the authors’ knowledge
of the local situation, the sample reflects the composition of the
3.2. Sample and data collection workforce in Korean deluxe tourist hotels fairly well.

The current study is based on the data from a survey executed in
Korean luxury hotels in four major metropolitan cities and a resort
island, including Seoul, Deagu, Kyungju, and Jeju. Considering the
total number of five star hotels in each city, a proportionate random
sampling was first applied to select participating hotels. Twenty-
two hotels were selected and full-time employees of each hotel
in room (front desk and housekeeping) and F&B divisions were
approached to participate in the survey. A small gift was given
to each participant upon completion of the survey questionnaire,
which required approximately 5–10 min to complete all close-
ended answers. A total of 450 questionnaires were distributed and
433 were returned. Among those 433 surveys, 31 unusable answers
were excluded, leaving 399 usable answers for this study.
3.3. Data analysis methods and procedures
The research model shown in Fig. 1 was analyzed primarily
using a structural equation modeling (SEM), supported by AMOS
7.0 (Arbuckle, 2006). Since a large portion of the workplace stres-
sor (i.e., CRS, WERS, and JRS) items were derived from previous
studies in other fields, they were modified through in-depth inter-
views with hotel managers to ensure their relevance to hotel
workplaces. Thus, to assess the validity and reliability of the dimen-
sions of the workplace stressors, an EFA was first conducted.
Numerous researchers have proposed a two-stage model-building
process for applying a SEM (Anderson and Gerbing, 1988; Hair
et al., 2006; Hoyle, 1995), in which the measurement model via
 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
a confirmatory factory analysis (CFA) was tested before testing the
structural model. The measurement model specifies how hypothet-
ical constructs are measured in terms of the observed variables.
Furthermore, the structural model specifies structural relationships
among the latent variables and describes the amount of explained
variance (Anderson and Gerbing, 1988).
Most researchers agree that common method bias (CMB) is
a potentially serious threat, especially when using single infor-
mant surveys (Lindell and Brandt, 2000; MacKenzie et al., 1999;
Podsakoff et al., 2003). In this study, all constructs were surveyed
using a single informant (i.e., hotel frontline employee). In the
measurement model analysis stage, we thus used a confirmatory
factor-analytic approach to Harman one-factor test (Harman, 1976;
Podsakoff and Organ, 1986) to determine empirically whether or
not CMB threatened the interpretation of our results. This approach
is prevalent among the other studies (e.g., Brockman and Morgan,
2006; Karatepe et al., 2010).
Next, to examine the moderating effect of OL (i.e. RFL vs.
OSL) between workplace stressors and NA based on a multigroup
approach, OL differences were tested by statistically comparing the
corresponding path coefficients in the structural model (Ahuja and
Thatcher, 2005; Keil et al., 2000).
Finally, to investigate the mediating role of NA on the relation-
ship between workplace stressors and EE, both the Aroian version
of the Sobel test suggested in Baron and Kenny’s (1986) study
and Preacher and Hayes’s (2004) bias-corrected (BC) bootstrap-
ping method were used. In particular, the statistical significance
of the coefficients of the measurement and structural model was
determined by a bootstrap resampling procedure (200 subsamples
were randomly generated). Developed by Efron (1987), this is now
a commonly used resampling method.
4. Findings
4.1. Exploratory factor analysis and validity and reliability of
workplace stressors
An eigenvalue of 1.0 was used for factor inclusion, and a fac-
tor loading of 0.40 was used as the benchmark to include items in
each factor. The appropriateness of an EFA was first determined
by the Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy
and Bartlett’s test of sphericity. The KMO was 0.864 and Bartlett’s
test of sphericity was significant at a level of p < 0.001, which justi-
fied the use of an EFA. Subsequently, the data were subjected to
EFA using a Varimax orthogonal rotation to reduce the number
of scale items. As a result (Table 2), three factors were derived
from the 14 workplace stressor scale items, which explained
62.092% of the variance. Based on the information of factor load-
ings and contents of the factors, the factors were labeled as JRS
(eigenvalue = 5.357; explained variance = 38.264; ˛ = 0.855), WERS
(eigenvalue = 0.936; explained variance = 13.829%; ˛ = 0.813), and
CRS (eigenvalue = 1.401; explained variance = 9.998%; ˛ = 0.802).
Thus, these three dimensions of workplace stressors (JRS, WERS,
and CRS) were employed as exogenous latent variables in the SEM
procedures.
4.2. Measurement model estimation
The strength of the measurement model can be demonstrated
through measures of convergent and discriminant validity. As
shown in Table 3, the standardized factor loadings were all sig-
nificant (t-value > 1.96) (Hair et al., 2006). It should be noted that
one CRS item (“some customers ask us to do things they could do
by themselves”) had a low SMC (below 0.40) and was dropped due
to its weak contribution to the construct (Hair et al., 2006). The
1073
composite reliability (CR) values ranged from 0.813 (CRS) to 0.938
(CO); all were above the recommended level of 0.70 for a reli-
able construct (Fornell and Larcker, 1981). The average variance
extracted (AVE) for each construct was between 0.566 (JRS) and
0.766 (NA), exceeding the acceptable value of 0.50 (Fornell and
Larcker, 1981). Hence, the measurement scales used in this study
established convergent validity. In addition, to examine the dis-
criminant validity of the measurement model, AVE values for the
constructs were compared to the squared correlations among the
constructs (Fornell and Larcker, 1981). None of the squared cor-
relations surpassed the AVE (Table 4). The tests indicated that
discriminant validity was upheld for the measurement model.
To demonstrate a reasonable fit for the model, a number of mea-
sures were computed, including Q, goodness-of-fit index (GFI), root
mean square of approximation (RMSEA), and comparative fit index
(CFI) (Table 3). The Q assumed that values less than 3.0 to as high
as 5.0 were indicative of an acceptable fit (Carmines and McIver,
1981). The Q was 2.008, less than the recommended 5.0 threshold.
The GFI (0.905) and CFI (0.948) values exceeded the recommended
0.90 threshold (Hair et al., 2006). The RMSEA was 0.050, less than
the cutoff value of 0.08 (Bollen, 1989). Thus, all statistics supported
satisfactory measurement quality given the number of indicators.
4.3. Common method bias checking
All constructs were surveyed using a single informant in this
study. Thus, we used a procedure to determine empirically whether
or not CMB threatened the interpretation of our results: a confirma-
tory factor-analytic approach to Harman one-factor test (Harman,
1976; Podsakoff and Organ, 1986). The rationale behind this
approach is that if CMB poses a serious threat to the analy-
sis and interpretation of the data, a single latent factor would
account for all manifest variables or one general factor would
account for the majority of the covariance among the measures. A
worse fit for the one-factor model, in comparison to a multi-factor
model, would suggest that CMB does not pose a serious threat.
The one-factor model yielded a 2 = 1798.225 with 275 degrees
of freedom; Q = 6.539; GFI = 0.614; RMSEA = 0.116; and CFI = 0.662
(compared with the 2 = 520.000 with 260 degrees of freedom;
Q = 2.008; GFI = 0.905; RMSEA = 0.050; and CFI = 0.948 for the six-
factor model reported above). The fit is considerably worse for
the one-dimensional model than it is for the measurement model,
which indicates that CMB is not a serious threat in this study. The
results proved that there is no instrument bias. Overall, we can
conclude that CMB does not threaten the interpretation of our data
analysis.
4.4. Structural model estimation and hypotheses testing
The goodness of fit statistics from the SEM showed that the
model reasonably fits the data (Table 5 and Fig. 2). The overall
fit of the model appeared to be acceptable with 2 (244), Q, GFI,
RMSEA, and CFI of 490.768, 2.011, 0.907, 0.050, and 0.948, respec-
tively. The results of the multivariate test of the structural model
indicated that as a whole the model explains 55.6% of the variance
in NA. The explained variances for EE and CO were 31.5% and 10.5%,
respectively.
Regarding the relationships between workplace stressors and
NA, CRS ( CRS to NA = 0.162, p < 0.01), WERS ( WERS to NA = 0.137,
p < 0.05) and JRS ( JRS to NA = 0.567, p < 0.001) all appeared to have
significant positive influences on NA, supporting Hypotheses 1a,
1b and 1c. NA was significantly and positively related to EE
(ˇNA to EE = 0.562, p < 0.001). Hypothesis 2 was thus supported. Fur-
thermore, as expected, EE had a significant negative effect on CO
1074 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082

Table 2
Results of exploratory factor analysis of workplace stressors.

Factor and indicators Factors

1. JRS 2. WERS 3. CRS

Job-related stressor (JRS)

My work often interrupts my personal timea 0.787
My coworkers often uncooperative to me 0.774
My work schedules change frequently 0.757
We are compelled to be in unnecessary training programs occasionallya 0.750
My workload is excessively heavy 0.712
I must deal with an excessively large number of customers 0.598

Work environment-related stressor (WERS)

My workplace is poorly lighted 0.809
The temperature in my workplace is not suitable 0.800
I don’t have enough space to refresh myself 0.768
There is not enough rest space in my workplacea 0.708

Customer-related stressor (CRS)

Customers often complain about us 0.813
Customers often vent their bad mood out on us 0.812
Customer demands are often exorbitant 0.758
Some customers ask us to do things they could do by themselves 0.668

Eigenvalue 5.357 1.936 1.401

Explained variance (%) 38.264 13.829 9.998
Total explained variance (%) = 62.092
Cronbach’s ˛ 0.855 0.813 0.802
Cronbach’s ˛ of the 14 workplace stressor scale items = 0.873
Kaiser–Meyer–Olkin measure of sampling adequacy = 0.864
Bartlett’s test of sphericity (significance level) = 0.000

Note:
a
Items were developed in this study and other items were adopted from previous related studies and modified.

Table 3
Results of reliability and convergent validity test.

Constructs and indicators Standardized factor loadings SMCs AVE CR

Customer-related stressor 0.593 0.813

Customers’ demands are often exorbitant 0.689 0.475
Some customers ask us to do things they could do by themselvesa − −
Customers often complain about us 0.748 0.560
Customers often vent their bad mood out on us 0.818 0.669

Work environment-related stressor 0.568 0.839

There is not enough rest space in my workplace 0.687 0.472
The temperature in my workplace is not suitable 0.774 0.599
My workplace is poorly lighted 0.785 0.616
I don’t have enough space to refresh myself 0.643 0.413

Job-related stressor 0.566 0.867

I must deal with an excessively large number of customers 0.634 0.402
My workload is excessively heavy 0.802 0.643
My work schedules change frequently 0.771 0.594
My coworkers often uncooperative to me 0.715 0.511
My work often interrupts my personal time 0.759 0.576
We are compelled to be in unnecessary training programs occasionally 0.637 0.406

Negative affectivity 0.766 0.907

Minor setbacks sometimes irritate me too much 0.858 0.736
I get often irritated at little annoyances 0.910 0.828
There are days when I am “on edge” all of the time 0.832 0.692

Emotional exhaustion 0.581 0.846

I feel emotionally drained from my work 0.718 0.526
I feel used up at the end of the workday 0.790 0.524
I feel frustrated by my job 0.722 0.521
Working with people all day is really a strain for me 0.697 0.486

Customer orientation 0.753 0.938

Every customer is important to me 0.710 0.504
I thrive on giving individual attention to each customer 0.796 0.634
I try to provide various information to the customers for their needs 0.848 0.719
I try to determine how I can best help the customer solve his/her problem 0.771 0.594
I generally know what customers want before they ask 0.763 0.582

Model fit measures
2 (260) = 520.000; Q = 2.008; GFI = 0.905; RMSEA = 0.050; CFI = 0.948

Note: All standardized factor loadings are significant at p < 0.001 level.
a
Item was deleted after convergent validity test.
 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082 1075

(ˇEE to CO = −0.324, p < 0.01). Hence, Hypothesis 3 was also sup-

−0.140*** (0.020)
−0.050NS (0.003)

−0.170** (0.029)
−0.141** (0.020)
−0.126* (0.016)
ported.

4.5. Testing the moderating role of organizational level and

0.753
hypotheses

6
In order to test the moderating effect of OL between work-

(0.261)
(0.215)
(0.104)
(0.106)
place stressors and NA based on a multigroup approach, this study
divided the sample data set into two subgroups: an RFL group with
0.323***
0.326***
0.511***
0.464***
326 cases and an OSL group with 73 cases. The workplace stressors

0.581
– CRS, WERS, and JRS – were postulated to have stronger positive
5

effects on NA for RFLs than for OSLs.

As can be seen in Table 5 and Fig. 3, hypotheses on OL differ-
0.414*** (0.171)
0.412*** (0.170)
0.690*** (0.476)

ences (Hypotheses 4a–4c) were tested by statistically comparing

Notes: The zero-order correlations and the calculated values of the squared correlations (brackets) among all constructs are presented in the upper offdiagonal. NS, non significant. the corresponding path coefficients in the structural model (Ahuja
and Thatcher, 2005; Keil et al., 2000). The statistical compari-
0.766

son was conducted by applying a procedure originally proposed

4

by Chin (1994) (Appendix A.1). Hypothesis 4a was tested by sta-

tistically comparing the path coefficient from CRS to NA in the
0.446*** (0.199)
0.434*** (0.188)

structural model for the RFL data set ( CRS to NA = 0.189) to the
corresponding path coefficient in the structural model for the
Zero-order correlations and discriminant validity

OSL data set ( CRS to NA = 0.063). The results indicate that the path
0.566

coefficient from CRS to NA for the RFL data set was significantly
3

stronger than the corresponding path coefficient for the OSL data
set (t = 9.001, p < 0.001). Likewise, Hypothesis 4b was tested by sta-
0.209*** (0.044)

tistically comparing the path coefficient from WERS to NA in the

structural model for the RFL data set ( WERS to NA = 0.179) to the
0.568

corresponding path coefficient in the structural model for the OSL

data set ( WERS to NA = 0.072). The results indicate that the path coef-
2

ficient from WERS to NA for the RFL data set was significantly
stronger than the corresponding path coefficient for the OSL data
set (t = 8.917, p < 0.001). Finally, Hypothesis 4c was tested by sta-
tistically comparing the path coefficient from JRS to NA in the
structural model for the RFL data set ( JRS to NA = 0.547) to the cor-
0.593

responding path coefficient in the structural model for the OSL data
1

set ( JRS to NA = 0.494). The results indicate that the path coefficient
from JRS to NA for the RFL data set was significantly stronger than
the corresponding path coefficient for the OSL data set (t = 26.514,
p < 0.001). As expected, RFL employees revealed a stronger rela-
tionship between CRS/WERS/JRS and NA compared to the OSL
employees; Hypotheses 4a–4c were thus supported.
0.792
0.747
0.879

0.591
0.801

0.740
Results of descriptive statistics, zero-order correlations, and discriminant validity test.

SD

4.6. Testing the mediating role of negative affectivity and an

Descriptive statistics

alternative model

In order to test the mediating role of NA on the relation-

ship between workplace stressors (i.e., CRS, WERS, and JRS) and
EE, direct paths from workplace stressors to EE were conducted
3.471
2.951
2.953
2.939
3.678
3.041

considering previous research. Zapf (2002) delineated that NA

M

exacerbated by EL has a direct positive effect on EE and depersonal-

ization. NA is evoked by stress factors, especially when employees
are involved more with surface acting and it tends to lead to EE and
job dissatisfaction, and other negative outcomes (Brotheridge and
Grandey, 2002; Brotheridge and Lee, 2003; Dallimore et al., 2007).
2. Work environment-related stressor

To test the mediating role of NA on the relationship between

workplace stressors and EE, both the Aroian version of the Sobel
test (Baron and Kenny, 1986) (Appendix A.2) and the BC bootstrap-
1. Customer-related stressor

ping method (Preacher and Hayes, 2004) were used. This combined
5. Emotional exhaustion

approach to mediation analysis was found to be more appropriate

6. Customer orientation
3. Job-related stressor
4. Negative affectivity

than traditional single mediation analysis approaches (MacKinnon

et al., 2002). The Aroian version of the Sobel test presents the level
of significance of the indirect effect of the independent variable on
p < 0.001.

the dependent variable through a mediator.

p < 0.05.
p < 0.01.
Construct

Estimates of the true indirect effect and its BC 95% bootstrap

Table 4

confidence interval (CI) are presented in Table 6. Both the Aroian

*

**

***

version of the Sobel test and BC bootstrapping method revealed

1076 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082

H1a:
Customer- 0.162 (3.035)**
related
Stressor

R2 = 0.556 R2 = 0.315 R2 = 0.105

H1b: H2: H3:
Work 0.137 (2.473)* 0.562 (10.004)*** −0.324 (−4.926)***
Environme Negative Emotional Customer
nt-related Affectivity Exhaustion Orientation
Stressor

Model fit measures

χ2 (244) = 490.768
Q = 2.011
GFI = 0.907
Job-related RMSEA = 0.050
Stressor H1c: CFI = 0.948
0.567 (8.392)***
Note: *p < 0.05, **p < 0.01, ***p < 0.001.

Fig. 2. Results of structural model estimation and hypotheses test.

that NA significantly mediated the effect of CRS ( CRS→NA→EE ; 0.030, 95% bootstrap CI = 0.012L CI , 0.183UL CI ; Z = 2.275, p < 0.05) on EE.
p < 0.01; 95% bootstrap CI = 0.005L CI , 0.080UL CI ; Z = 2.113, p < 0.05), Specifically, the direct effect of JRS ( JRS to EE ; 0.373, p < 0.001) on EE
WERS ( WERS→NA→EE ; 0.025, p < 0.05; 95% bootstrap CI = 0.004L CI , was positively significant. Therefore, this result indicates that NA
0.076UL CI ; Z = 2.034, p < 0.05), and JRS ( JRS→NA→EE ; 0.103, p < 0.05; has a partial mediating role on the relationship between JRS and EE.

Table 5
Results of main and moderating effect test.

Main effect Results

Standardized path coefficients t-values

Hypothesized relationships
Customer-related stressor → negative affectivity (H1a; +) 0.162 3.035** Accepted
Work environment-related stressor → negative affectivity (H1b; +) 0.137 2.473* Accepted
Job-related stressor → negative affectivity (H1c; +) 0.567 8.392*** Accepted
Negative affectivity → emotional exhaustion (H2; +) 0.562 10.004*** Accepted
Emotional exhaustion → customer orientation (H3; −) −0.324 −4.926** Accepted

Explanatory power
Negative affectivity 0.556 (55.6%)
Emotional exhaustion 0.315 (31.5%)
Customer orientation 0.105 (10.5%)
Model fit measures
2 (244) 490.768
Q 2.011
GFI 0.907
RMSEA 0.050
CFI 0.948

Moderating effect Results

RFL (n = 326) OSL (n = 73) t-Value

difference test

Standardized t-Values Standardized t-Values

path path
coefficients coefficients

Hypothesized relationships
Customer-related stressor → negative affectivity (H4a; +) 0.189 3.361*** 0.063 0.337NS 9.001*** Accepted
Work environment-related stressor → negative affectivity (H4b; +) 0.179 2.980** 0.072 0.470NS 8.917*** Accepted
Job-related stressor → negative affectivity (H4c; +) 0.547 7.651*** 0.494 3.001** 26.514*** Accepted

Model fit measures
2 (244) 437.485 378.403
Q 1.793 1.551
GFI 0.899 0.733
RMSEA 0.049 0.087
CFI 0.952 0.831

Note: RFL, rank & file level; OSL, over supervisor level; and NS, non significant.
*
p < 0.05.
**
p < 0.01.
***
p < 0.001.
 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082 1077

H4a: H4a:
Customer- 0.189 (3.361)*** 0.063 (0.337)NS
Customer-
related related
Stressor Stressor

R2 = 0.600 R2 = 0.356
H4b: H4b:
Work 0.179 (2.980)** Work 0.072 (0.470)NS
Environme Negative Environme Negative
nt-related Affectivity nt-related Affectivity
Stressor Stressor

Job-related Job-related
Stressor H4c: Stressor H4c:
0.547 (7.651)*** 0.494 (3.001)**

Model fit measures Model fit measures

χ2 (244) = 437.485 χ2 (244) = 378.403
Q = 1.793 Q = 1.551
GFI = 0.899 GFI = 0.733
RMSEA = 0.049 RMSEA = 0.087
CFI = 0.952 CFI = 0.831
RFL data set (n = 326) OSL data set (n = 73)
** ***
Notes: RFL, rank & file level; OSL, over supervisor level; and NS, non significant. p < 0.01, p < 0.001.

Fig. 3. Comparison of parameter estimates in the rank and file level and over supervisor level data set.

However, the direct effects of CRS ( CRS to EE ; 0.084, NS) and WERS
( WERS to EE ; 0.087, NS) on EE turned out to be insignificant but pos-
itive. These results imply that NA fully mediates the relationships
between CRS/WERS and EE.
The model testing and comparison should focus on assessing
model fit and comparing the fit of competing models (Kelloway,
1998). As both the proposed model [2 (244) = 490.768; Q = 2.011;
GFI = 0.907; RMSEA = 0.050; CFI = 0.948] and the competing
model [2 (241) = 480.313; Q = 1.993; GFI = 0.909; RMSEA = 0.049;
CFI = 0.951] fit the original data appropriately (Figs. 2 and 4), the
two models were compared for model fit, path coefficients, and
explanatory power (R2 ) (T. Kim et al., 2009). A 2 difference test
was employed to determine if one structure performs better than
the other (Anderson and Gerbing, 1988), because the proposed
model is nested within the alternative model. The result shows that
the alternative model with direct paths from workplace stressors
to EE has a slightly better overall model fit than the proposed
model. The difference between the two models was significant
at p < 0.05 (2 = 10.455, df = 3). In terms of path coefficients
significance, all path coefficients in the proposed model were
statistically significant and supporting all the hypothesized rela-
tionships. However, in the alternative model, two (CRS to EE; WERS
to EE) of eight path coefficients were statistically insignificant
(Figs. 2 and 4). Finally, explanatory power was used to determine
which model was superior in predicting hotel frontline employees’
NA, EE, and CO. In the explanatory power of the alternative model,
NA variable did provide a better prediction of EE; in this case, the
R2 increased relative to the proposed model [the proposed model:
R2 (EE) = 0.315 (31.5%); the alternative model: R2 (EE) = 0.389
(38.9%)]. However, in the explanatory power of the alternative
model, CRS, WERS, and JRS variables did not provide a better
prediction of NA; the R2 decreased relative to the proposed model
[the proposed model: R2 (NA) = 0.556 (55.6%); the alternative
model: R2 (NA) = 0.530 (53%)]. Also, in the explanatory power
of the alternative model, EE variable did not provide a better
prediction of CO; the R2 decreased relative to the proposed model
[the proposed model: R2 (CO) = 0.105 (10.5%); the alternative
model: R2 (CO) = 0.102 (10.2%)]. In summary, the models shown in

Table 6
Results of mediating role estimation of negative affectivity between workplace stressors and emotional exhaustion: Aroian version of the Sobel test and BC bootstrapping
test.

Paths of mediating role Indirect effects 95% bootstrap CIs Z-values Mediating roles

LL CIs UL CIs

Customer related stressor → negative affectivity → emotional exhaustion 0.030** 0.005 0.080 2.113* Full mediator
Work environment-related stressor → negative affectivity → emotional exhaustion 0.025* 0.004 0.076 2.034* Full mediator
Job-related stressor → negative affectivity → emotional exhaustion 0.103* 0.012 0.183 2.275* Partial mediator

Notes: Because the relationship between job-related stressor and emotional exhaustion is significant and positive (Fig. 4), among the relationship of job-related
stressor → negative affectivity → emotional exhaustion, negative affectivity is a partial mediator; because the relationships between customer-related stressor/work
environment-related stressor and emotional exhaustion are insignificant positive (Fig. 4), among the relationships of customer-related stressor/work environment-related
stressor → negative affectivity → emotional exhaustion, negative affectivity is a full mediator. BC, bias-corrected; CI, confidence interval; LL, lower level; and UL, upper level.
*
p < 0.05.
**
p < 0.01.
1078 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
H1a:
Customer- 0.161 (2.955)**
related
Stressor 0.084 (1.296)NS
R2 = 0.530
0.087 (1.324)NS
H1b:
Work 0.134 (2.383)*
Environme Negative
nt-related Affectivity
Stressor
H1c:
0.552 (8.109)***
Job-related 0.373 (4.189)***
Stressor
Notes: For testing mediating role (full or partial mediator) of negative affectivity between workplace stressors (i.e., customer-related, work
environment related, and job-related stressor) and emotional exhaustion, the direct paths from workplace stressors to emotional exhaustion were
added. NS, non significant. *p < 0.05, **p < 0.01, ***p < 0.001.
Fig. 4. Results of structural model estimation of mediating role of negative affectivity.
Figs. 2 and 4 can be ordered in a nested sequence (nested model) to
compare model fit, parameter estimates, and exploratory power.
The differences in 2 and degrees of freedom (df), fit statistics,
as well as exploratory power indicate that the proposed model is
slightly superior to the alternative model.
5. Conclusions
5.1. Discussion
In this study the workplace stress factors that affect NA among
hotel frontline employees were identified and categorized into
three dimensions of CRS, WERS, and JRS, which are significant
antecedents of NA. More importantly, the effect on NA of each
factor was empirically tested and compared. As hypothesized, all
of the workplace stressors, CRS, WERS, and JRS, had significant
effects on NA among hotel frontline employees. These findings
can be explained based on Brymer et al.’s (1991) research, which
argued that hotel employees are involved with high levels of EL
and tend to experience more strain as a consequence of workplace
stress. Those stress factors collectively explained more than 55% of
the total variance of NA (Fig. 2) within the sample of this study.
The findings are consistent with previous studies (e.g., Ashforth
and Humphrey, 1993; Chen and Spector, 1991; Fullerton and Punj,
2004; Morris and Feldman, 1996; Sutton and Rafaeli, 1988), where
workplace stress factors are related to negative emotional and
behavioral outcomes. In this study the relative effect of each work-
place stress factor on NA was also compared. JRS turned out to
be the most influential factor affecting respondents’ NA; irregular
work schedule, unexpected calls, excessive workload, and force-
ful participation in training programs caused the most NA among
the respondents. Having to deal with overly demanding customers
was another stressor that increased hotel employees’ NA in service
encounters. Interestingly, CRS exerted less influence on NA com-
pared with JRS. Although significantly related to NA, WERS was the
least significant factor affecting NA. Given that the sample is from
five-star and above hotels, the employees may be given appropriate
physical facilities for their needs. Further, the respondents may
R2 = 0.389 R2 = 0.102
H2: H3:
0.187 (2.374)* −0.319 (−4.716)***
Emotional Customer
Exhaustion Orientation
Model fit measures
χ2 (241) = 480.313
Q = 1.993
GFI = 0.909
RMSEA = 0.049
CFI = 0.951
have sufficient training to value the business customers bring and
appreciate their guests in any situation. As found repeatedly in pre-
vious studies (e.g., Chen and Spector, 1991; Morris and Feldman,
1996), job ambiguity, work overloads, and irregular work sched-
ules seem to create more negative feelings, emotions, and attitudes
than any other factor.
Another important finding of this study is the moderating effect
of OL on the workplace stressors–NA relationship. The quality of
the relationship was explored by incorporating the OL effect in the
model as a qualifier of the workplace stressors–NA relationship.
The results show that all three dimensions of workplace stressors,
CRS, WERS, and JRS showed statistically significant effects among
RFL employees, whereas only JRS turned out to be significant for
OSL employees (Table 5 and Fig. 3). A formal test to compare the
corresponding path coefficients in the structural model confirmed
statistically significant differences across the two groups in all three
pairs of comparisons (Table 5 and Fig. 3). This finding supports pre-
vious work in OB and management fields (Ashforth and Humphrey,
1993; Chen and Spector, 1991; Morris and Feldman, 1996; Rafaeli
and Sutton, 1987; Sutton and Rafaeli, 1988). Given the nature of
the hotel industry, stressors in all three areas seem to impede the
coping process for lower level employees because they have less
authority and control over stressful conditions and limited coping
resources (Fisher and Gitelson, 1983; Hamner and Tosi, 1974).
As for the links in the workplace stressor–strain model of this
study, hotel frontline employees’ EE was hypothesized to have
a reverse relationship with CO. The empirical result indicated a
significant reverse relationship between employees’ EE and CO.
This result implies that when hotel frontline employees experi-
ence EE, they are less likely to have customer oriented attitudes
and behaviors. Accordingly, their active interaction with customers,
genuine interest in customers’ needs, willingness to provide infor-
mation to customers, and efforts to solve customers’ problems
decreased.
Finally, this study examined the direct and indirect effects in
an effort to gain insights into hotel employees’ EE process. Turning
to mediating role of NA between three workplace stressors (i.e.,
CRS, WERS, and JRS) and EE, our results indicated that NA fully
mediates the relationships between CRS, WERS and EE: when NA
 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
effect was removed from the relationship between CRS, WER and
EE, the effects from CRS and WER on EE turned out to be completely
insignificant. JRS, however, had both a significant positive direct
effect on EE and an indirect effect on EE via NA. Thus, NA is par-
tially mediated by the relationship between JRS and EE. This result
confirms the salient feature of NA as the link between employees’
perceived JRS and EE in hotels.
What we learned from this empirical study is that managing
hotel frontline employees’ EL is foundational for marketing success.
Hotel employees are exposed to heavy EL on a daily basis. Managing
their inner emotions, not just their surface behaviors, is a critical
HRM issue because it is ultimately tied to the long-term market-
ing goal: making lasting relationships with valued customers. The
complexity of the structural relationships among the variables in
stressor–strain relationship and their consequences has not been
fully disentangled in this one cross-sectional study. However, this
study makes a meaningful contribution to the existing body of
knowledge on EL in the hospitality industry. First, we identified and
categorized stress inducing factors and were able to compare the
relative magnitude of each effect on NA. Second, the effect of NA as
a mediator between workplace stress factors and strain outcomes
was also tested. Third, by postulating CO as a key outcome variable,
we incorporated HRM issues and marketing concerns in a single
model. Last, but not the least, we tested the moderating effect of
OL among hotel frontline employees, which helped us understand
why and how we should strategically deal with the EL of employees.
Although this study only used two groups of employees, RFL and
OSL, they showed distinctly different emotional reactions to work-
place stressors. These results imply many HRM related questions
and open new venues to explore in the future.
5.2. Managerial implications
The results provide several practical implications for the hotel
industry in terms of key HRM concerns, including training, team
solidarity, compensation, and coping and employee retention. First,
providing training programs to develop the necessary skills to per-
form EL and avoid depersonalization in a service encounter may
require more concerted efforts by organizations. Also, training pro-
grams that include empowerment practices are an effective way to
help frontline employees deal with customer complaints with more
confidence and increased satisfaction, and thus leading to fewer
emotional gaps and burnouts. One thing HR managers should con-
sider based on the results of this study is implementing separate
training programs for RFL and OSL employees.
With regards to training employees, it is critical for both employ-
ees and organizations to learn how to deal with situations where
customers believe the golden rule that “customers are always
right.” In the hospitality industry, the focal point for most training
programs is on the customer’s feelings. However, very few training
programs are designed to discuss how service employees feel. As
employees accumulate different experiences on the job, it is equally
important to openly discuss their feelings and frustration. This type
of open discussion delivers a message to employees that the com-
pany acknowledges their emotional contributions. This positive
feedback can motivate employees to increase their productivity and
commitment to their jobs and organizations. By implementing this
type of training, employees can develop the ability to effectively
vent their anxiety and NA before it accumulates. Another focus of
employee training is on how to control employee NA in service
encounters. One example is the deep acting technique. Deep act-
ing can be a shield against being susceptible to stressor effects.
By integrating deep acting into employee training programs, the
internal processes of emotional management are carefully regu-
lated and external emotional expressions seem more genuine and
spontaneous (Hochschild, 1983).
1079
Frontline employees are not the only people performing EL.
Managers do a great deal of EL as well. They monitor, control, and
direct the EL employees perform for customers. Hochschild (1983)
commented that “What is offstage for the frontline employees is
on-stage for the supervisor” (p. 118). It is the manager’s responsi-
bility to ensure that their employees are in the best condition to
perform. Accordingly, organizations need to train their managers
how to perform EL when interacting with their employees. How to
provide emotional support, handle employee complaints, provide
feedback, and give constructive opinions should be integrated into
formal management training programs.
Second, from a team solidarity perspective, quality service is
accomplished through the cooperation of a team. Team solidarity
can improve morale and thus improve service. Building a sense
of teamwork not only helps in the service delivery process, but
also nurtures a positive mood among team players. Third, from a
compensation perspective, although EL is crucial in determining
customer satisfaction and affects business performance, employ-
ees’ emotional contributions have not been fully acknowledged.
While industries tend to compensate for mental and physical labor,
most of the EL performed in the hospitality industry does not
receive due compensation in monetary or non-monetary forms.
When service employees successfully deep act to provide quality
service, they need to be not only recognized, but also rewarded.
Finally, from a coping-mechanism and employee retention per-
spective, stressor factors in the hospitality industry cannot be
fully controlled. Thus, hotel management should implement prac-
tical coping mechanisms. First, management must define a policy
regarding customers’ rights and the maximum level of tolerance
for offensive customers (see Bailey, 1994). Also, the policy must be
accompanied by written procedures. In addition, empowerment
appears to be an effective measure in managing guest miscon-
duct and sending a powerful signal to employees that the hotel
management trusts them.
As the results show, frontline employees in the hotel indus-
try tend to have weaker stressor–strain relationship as they move
up the organizational hierarchy. This also implies that employee
turnover is detrimental not only in terms of organizational stability
but also for successful service delivery, as employees in lower ranks
(RFL) tend to be strained more by on the job stressors than their
more experienced counterparts (OSL). It is also critical to imple-
ment predictable and regular work scheduling practices, given that
JRS is the biggest work related stressor. If irregular work schedules
are inevitable, employees must have notice in advance so that they
can readjust their personal schedules. Excessive workloads, role
conflict, and role ambiguity also must be properly controlled. Man-
agerial practices that accommodate personal needs of employees
can greatly help reduce employee turnover due to EE. Most of all, it
should be noted that for the most effective results these solutions
should be implemented in tandem.
As the findings clearly indicate, all these suggestions to man-
age employees’ feelings are foundational for quality service and its
delivery, since frontline employees of hotels are the most important
element of marketing strategy (Singh, 2000). As evident in recent
empirical studies (Babakus et al., 2009), frontline employees’ CO
is directly linked to their job performance and also acts as a buffer
against EE. Therefore, the quality of their emotional work is inextri-
cably tied to the source of competitive advantage over competitors.
We suggest that future research should empirically examine the
diverse roles of CO in the context of hotel employees’ EL from a
marketing perspective.
5.3. Limitations and future research
We analyzed the structural relationship of workplace
stressor–strain and its consequence in the context of EL in
1080 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
the hotel industry. This study is based on cross-sectional survey
data. This approach, however, limits our grasp on the sheer com-
plexity of the relationship. Diverse research methods, including
observation and interpretation, experiment design, and individual
or focus group interviews may enhance our knowledge on the
dynamics of causality.
Three items modified from Watson and Clark (1984) were used
to measure the concept of NA in this study. Even though a succinct
sub-scale of the original NA scale did not fail to capture the core
concept of NA, it must be noted that a few items in the original
full scale showed low communality and double loaded. This may
stemmed from cultural or semantic differences. This issue regard-
ing NA measurement scale deserves a further investigation in the
future research.
In our study, the type of EL was not manipulated. EL has
two dimensions: surface acting and deep acting (Hochschild,
1979). Interestingly, while surface acting tends to lead to neg-
ative emotional/behavioral outcomes, deep acting can lead to
positive outcomes such as feelings of personal accomplish-
ment (Brotheridge and Lee, 2003), less burnout (Grandey, 2003),
and greater service performance (Grandey, 2003; Totterdell and
Holman, 2003). These contrasting effects mark the distinction
between emotional dissonance and emotional congruence, which
could occur simultaneously depending on the type of EL. There
must be a sub-mechanism not incorporated in our current study
model that rules the relationship between types of acting and emo-
tional dissonance/congruence outcomes. This mechanism holds
much promise for future empirical investigations.
Another drawback of the model in the current study is that
it did not include individual traits, coping strategies, and coping
resources for their potentially powerful mediating or moderating
effects on EE (Chen and Spector, 1991). Experience in EL contexts is
based on the personal assessment of stressors and their individual
strategies to prevent strains (Lazarus, 1993). Therefore, incorporat-
ing EE variable will provide a more thorough understanding and a
fruitful venue for future research.
Acknowledgements
The authors would like to thank the constructive comments
offered by the anonymous reviewers of this paper. This work was
supported by a grant from the Kyung Hee University in 2011 (KHU-
20110096).
Appendix A.
A.1. Procedure and formula for moderating role testing of OL
The statistical comparison between RFL and OSL was conducted
by applying a procedure originally proposed by Chin (1994) as fol-
lows:
PCR − PC0
t-Value =  (1)
1 1
Spooled × NR
+ N0
From Eq. (1), Spooled is expressed as:
 Brymer, R.A., Perrewe, P.L., Johns, T.R., 1991. Managerial stress in the hotel industry.
NR − 1 N0 − 1
Spooled = × SER2 + × SE02
NR + N0 − 2 NR + N0 − 2
where Spooled , pooled estimator for the variance; t, t-statistic with
NR + N0 − 2 degrees of freedom; NR , sample size of data set for RFL;
N0 , sample size of data set for OSL; SER , standard error of path in
structural model of RFL; SE0 , standard error of path in structural
model of OSL; PCR , path coefficient in structural model of RFL; PC0 ,
path coefficient of path in structural model of OSL.
A.2. The Aroian version of the Sobel test for mediating role testing
of NA
The Aroian version of the Sobel test (Baron and Kenny, 1986) is
represented as follows:
a×b
Z-Value =  (3)
b2 × sa2 + a2 × sb2 + sa2 × sb2
where a and b are unstandardized path coefficients of paths and sa
and sb are associated standard errors.
References
Agho, A.O., Price, J.L., Mueller, C.W., 1992. Discriminant validity of measures of job
satisfaction, positive affectivity and negative affectivity. Journal of Occupational
and Organisational Psychology 65 (3), 185–196.
Agho, A.O., Mueller, C.W., Price, J.L., 1993. Determinants of employee job satisfaction:
an empirical test of a causal model. Human Relations 46 (8), 1007–1027.
Ahuja, M.K., Thatcher, J.B., 2005. Moving beyond intentions and toward the theory of
trying: effects of work environment and gender on post-adoption information
technology use. MIS Quarterly 29 (3), 427–459.
Anderson, J.C., Gerbing, D.W., 1988. Structural equation modeling in practice: a
review and recommended two-step approach. Psychological Bulletin 103 (3),
411–423.
Anderson, S.E., Coffey, B.S., Byerly, R.T., 2002. Formal organizational initiatives and
informal workplace practices: links to work-family conflict and job-related out-
comes. Journal of Management 28 (6), 787–810.
Arbuckle, J.L., 2006. AMOS 7.0 User’s Guide. SPSS, Chicago, IL.
Ashforth, B.E., Humphrey, R.H., 1993. Emotional labor in service roles: the influence
of identity. Academy of Management Review 18 (1), 88–115.
Babakus, E., Yavas, U., Karatepe, O.M., 2008. The effects of job demands, job resources
and intrinsic motivation on emotional exhaustion and turnover intentions: a
study in the Turkish hotel industry. International Journal of Hospitality and
Tourism Administration 9 (4), 384–404.
Babakus, E., Yavas, U., Ashill, N.J., 2009. The role of customer orientation as a moder-
ator of the job demand burnout–performance relationship: a surface-level trait
perspective. Journal of Retailing 85 (4), 480–492.
Babin, B.J., Boles, J.S., 1998. Employee behavior in a service environment: a model
and test of potential differences between men and women. Journal of Marketing
62 (2), 77–91.
Bagozzi, R.P., Gopinath, M., Nyer, P.U., 1999. The role of emotions in marketing.
Journal of the Academy of Marketing Science 27 (2), 184–206.
Bailey, D., 1994. Recovery from customer service shortfalls. Managing Service Qual-
ity 4 (6), 25–28.
Baron, R.M., Kenny, D.A., 1986. The moderator–mediator variable distinction in
social psychological research: conceptual, strategic, and statistical considera-
tions. Journal of Personality and Social Psychology 51 (6), 1173–1182.
Beehr, T.A., Newman, J.E., 1978. Job stress, employee heath, and organizational effec-
tiveness: a facet analysis, model, and literature review. Personnel Psychology 31
(4), 665–699.
Beehr, T.A., King, L.A., King, D.W., 1990. Social support and occupational stress:
talking to supervisors. Journal of Vocational Behavior 36 (1), 61–81.
Bitner, M., 1992. Servicescape: the impact of physical surroundings on customers
and employees. Journal of Marketing 56 (2), 57–71.
Bollen, K.A., 1989. Structural Equations with Latent Variables. Wiley, New York, NY.
Bradburn, N.M., 1969. The structure of psychological well-being. Aldine, Chicago, IL.
Brief, A.P., Burke, M.J., Atieh, J.M., Robinson, B.S., Webster, J., 1988. Should negative
affectivity remain an unmeasured variable in the study of job stress? Journal of
Applied Psychology 73 (2), 193–198.
Brockman, B.K., Morgan, R.M., 2006. The moderating effect of organizational cohe-
siveness in knowledge use and new product development. Journal of the
Academy of Marketing Science 34 (3), 295–307.
Brotheridge, C.M., Grandey, A.A., 2002. Emotional labor and burnout: comparing two
perspectives of people work. Journal of Vocational Behavior 60 (1), 17–39.
Brotheridge, C.M., Lee, R.T., 2003. Development and validation of the emotional labor
scale. Journal of Occupational and Organizational Psychology 76 (3), 365–379.
Brown, T.J., Mowen, J.C., Donavan, D.T., Licata, J.W., 2002. The customer orientation
of service workers: personality trait effects on self- and supervisor performance
ratings. Journal of Marketing Research 39 (1), 110–119.
International Journal of Hospitality Management 10 (1), 47–58.
Carmines, E., McIver, J.M., 1981. Analyzing models with unobserved variables. In:
(2)
Bohrnstedt, G., Borgath, E. (Eds.), Social Measurement: Current Issue. Sage, Lon-
don.
Chen, P.Y., Spector, P.E., 1991. Negative affectivity as the underlying cause of cor-
relations between stressors and strains. Journal of Applied Psychology 76 (3),
398–407.
Chen, S., Westman, M., Eden, D., 2009. Impact of enhanced resources on anticipatory
stress and adjustment to new information technology: a field-experimental test
of conservation of resources theory. Journal of Occupational Health Psychology
14 (3), 219–230.
Chin, W.W., 1994. PLS-Graph Manual Version 2.7. University of Calgary, Calgary, AB.
 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
Cohen, S., Willis, T.A., 1985. Stress, social support, and the buffering hypothesis.
Psychological Bulletin 98 (2), 310–357.
Cohen, S., Edwards, J.R., 1989. Personality characteristics as moderators of the rela-
tionship between stress and disorder. In: Meufeld, R.W.J. (Ed.), Advances in the
Investigation of Psychological Stress. Wiley, New York, NY.
Cordes, C.L., Dougherty, T.W., 1993. A review and an integration of research on job
burnout. Academy of Management Review 18 (4), 621–656.
Cropanzano, R., Rupp, D.E., Byrne, Z.S., 2003. The relationship of emotional
exhaustion to work attitudes, job performance, and organizational citizenship
behaviors. Journal of Applied Psychology 88 (1), 160–169.
Dallimore, K.S., Sparks, B.A., Butcher, K., 2007. The influence of angry customer out-
bursts on service providers’ facial displays and affective states. Journal of Service
Research 10 (1), 78–92.
Diener, E., Emmons, R.A., 1984. The independence of positive and negative affect.
Journal of Personality and Social Psychology 47 (5), 1105–1117.
Dormann, C., Zapf, D., 2004. Customer-related social stressors and burnout. Journal
of Occupational Health Psychology 9 (1), 61–82.
Efron, B., 1987. Better bootstrap confidence intervals (with discussion). Journal of
the American Statistical Association 82 (397), 171–200.
Fisher, C.D., Gitelson, R., 1983. A meta-analysis of the correlates of role conflict and
ambiguity. Journal of Applied Psychology 68 (2), 320–333.
Fornell, C., Larcker, D., 1981. Evaluating structural equation models with unob-
servable and measurement error. Journal of Marketing Research 18 (1),
39–50.
Fullerton, R.A., Punj, G., 2004. Repercussions of promoting an ideology of consump-
tion: consumer misbehavior. Journal of Business Research 57 (11), 1239–1249.
Furnham, A., Hughes, K., 1999. Individual difference correlates of night work and
shift work relation. Personality and Individual Differences 26 (5), 941–959.
Gaines, J., Jermier, J.M., 1983. Emotional exhaustion in a high stress organization.
Academy of Management 26 (4), 567–586.
George, J.M., 1989. Mood and absence. Journal of Applied Psychology 74 (2),
317–324.
Grandey, A.A., 2003. When the show must go on: surface acting and deep acting as
determinants of emotional exhaustion and peer-rated service delivery. Academy
of Management Journal 46 (1), 86–96.
Grandey, A.A., Kern, J.H., Frone, M.R., 2007. Verbal abuse from outsiders versus
insiders: comparing frequency, impact on emotional exhaustion, and the role
of emotional labor. Journal of Occupational Health Psychology 12 (1), 63–79.
Hair Jr., J.F., Black, W.C., Babin, B.J., Anderson, R.E., Tatham, R.L., 2006. Multivariate
Data Analysis, sixth ed. Prentice Hall, Upper Saddle River, NJ.
Hamner, W.C., Tosi, H.L., 1974. Relationship of role conflict and role ambi-
guity to job involvement measures. Journal of Applied Psychology 59 (4),
497–499.
Harman, G., 1976. Practical reasoning. Review of Metaphysics 29, 431–463.
Hartline, M.D., Maxham III, J.G., McKee, D.O., 2000. Corridors of influence in
the dissemination of customer-oriented strategy to customer contact service
employees. Journal of Marketing 64 (2), 35–50.
Hobfoll, S.E., 1989. Conservation of resources: a new attempt at conceptualizing
stress. American Psychologist 44 (3), 513–524.
Hobfoll, S.E., 2001. The influence of culture, community, and the nested-self in the
stress process: advancing conservation of resources theory. Applied Psychology:
An International Review 50 (3), 337–421.
Hochschild, A.R., 1979. Emotion work, feeling rules, and social structure. American
Journal of Sociology 85 (3), 551–575.
Hochschild, A.R., 1983. The Managed Heart. University of California Press, Berkeley.
Hoffman, K., Ingram, T., 1991. Creating customer-oriented employees: the case in
home health care. Journal of Health Care Management 11 (2), 24–32.
Houkes, I., Janssen, P.P.M., De Jonge, J., Nijhuis, F.J.N., 2001. Work and individual
determinants of intrinsic work motivation, emotional exhaustion, and turnover
intention: a multi-sample analysis. International Journal of Stress Management
8 (4), 257–283.
Hoyle, R.H., 1995. Structural Equation Modeling. Sage, Thousand Oaks, CA.
Iverson, R.D., Olekalns, M., Erwin, P.J., 1998. Affectivity, organizational stressors,
and absenteeism: a causal model of burnout and its consequences. Journal of
Vocational Behavior 52 (1), 1–23.
Jamal, M., 2004. Burnout, stress, and health of employees on nonstandard work
schedules: a study of Canadian workers. Stress and Health 20 (3), 113–119.
Judge, T.A., Woolf, E.F., Hurst, C., 2009. Is emotional labor more difficult for some than
for others? A multilevel, experience-sampling study. Personnel Psychology 62
(1), 57–88.
Julian, C.C., 2008. Emotional dissonance and customer service: an exploratory study.
Service Marketing Quarterly 29 (3), 1–23.
Karatepe, O.M., 2006. The effects of selected antecedents on the service recovery per-
formance of frontline employees. The Service Industries Journal 26 (1), 39–57.
Karatepe, O.M., Uludag, O., 2008. Affectivity, conflicts in the work–family interface,
and hotel employee outcomes. International Journal of Hospitality Management
27 (1), 30–41.
Karatepe, O.M., Aleshinloye, K.D., 2009. Emotional dissonance and emotional
exhaustion among hotel employees in Nigeria. International Journal of Hospi-
tality Management 28 (3), 349–358.
Karatepe, O.M., Yorganci, I., Haktanir, M., 2009. Outcomes of customer verbal aggres-
sion among hotel employees. International Journal of Contemporary Hospitality
Management 21 (6), 713–733.
Karatepe, O.M., Yorganci, I., Haktanir, M., 2010. An investigation of the role of job
resources in mitigating customer-related social stressors and emotional exhaus-
tion. Service Marketing Quarterly 31 (1), 72–88.
1081
Karatepe, O.M., 2011. Do job resources moderate the effect of emotional dissonance
on burnout? International Journal of Contemporary Hospitality Management 23
(1), 44–65.
Keil, M., Tan, B.C.Y., Wei, K.-K., Saarinen, T., Tuunainen, V., Wassenaar, A., 2000. A
cross-cultural study on escalation of commitment behavior in software projects.
MIS Quarterly 24 (2), 299–324.
Kelloway, E.K., 1998. Using LISREL for Structural Equation Modeling: A Researcher’s
Guide. Sage, Thousand Oaks, CA.
Kennedy, K.N., Lassk, F.G., Goolsby, J.R., 2002. Customer mindset of organization-
wide employees. Journal of the Academy of Marketing Science 30 (2), 159–171.
Kim, B.P., Murrmann, S.K., Lee, G., 2009. Moderating effects of gender and organi-
zational level between role stress and job satisfaction among hotel employees.
International Journal of Hospitality Management 28 (4), 612–619.
Kim, T., Kim, W.G., Kim, H.-B., 2009. The effects of perceived justice on recovery sat-
isfaction, trust, worth-of-mouth, and revisit intention in upscale hotels. Tourism
Management 30 (1), 51–62.
Kim, K.-H., Kwon, J.-H., 2010. Effect of emotional dissonance and exhaustion on cus-
tomer orientation for hotel F&B employees. Journal of Foodservice Management
Society of Korea 13 (1), 215–231.
Kim, S.H., Suh, M.S., 2005. A study on the influencing and moderating factors of ser-
vice providers’ emotional dissonance in the service encounter. Journal of Korean
Marketing Association 20 (1), 111–145.
King, L.A., Emmons, R.A., 1990. Conflict over emotional expression: psychological
and physical correlates. Journal of Personality and Social Psychology 58 (5),
864–877.
Lazarus, R.S., 1993. Coping theory and research: past, present, and future. Psycho-
somatic Medicine 55 (3), 234–247.
Levin, I., Stokes, J., 1989. Dispositional approach to job satisfaction: role of negative
affectivity. Journal of Applied Psychology 74 (5), 752–758.
Licata, J.W., Mowen, J.C., Harris, E.G., Brown, T.J., 2003. On the trait antecedents and
outcomes of service worker job resourcefulness: a hierarchical model approach.
Journal of the Academy of Marketing Science 31 (3), 256–271.
Lindell, M.K., Brandt, C.J., 2000. Climate quality and consensus as mediators of
the relationship between organizational antecedents and outcomes. Journal of
Applied Psychology 85 (3), 331–348.
Liu, Y., Perrewe, P.L., Hochwarter, W.A., Kachmar, C.J., 2004. Dispositional
antecedents and consequences of emotional labor at work. Journal of Leadership
and Organizational Studies 10 (4), 12–25.
Loseke, D.R., Cahill, S.E., 1986. Actors in search of a character: student social workers’
quest for professional identity. Symbolic Interaction 9 (2), 245–258.
MacKenzie, S.B., Podsakoff, P.M., Paine, J.B., 1999. Do citizenship behaviors matter
more for managers than for salespeople? Academy of Marketing Science Journal
27 (4), 396–410.
MacKinnon, D.P., Lockwood, C.M., Hoffman, J.M., West, S.G., Sheets, V., 2002. A com-
parison of methods to test mediation and other intervening variable effects.
Psychological Methods 7 (1), 83–104.
Maslach, C., Jackson, S.E., 1981. The measurement of experienced burnout. Journal
of Occupational Behavior 2 (2), 99–113.
Maslach, C., 1982. Burnout: The Cost of Caring. Prentice-Hall, Englewood Cliffs, NJ.
Morris, J.A., Feldman, D.C., 1996. The dimensions, antecedents, and consequences of
emotional labor. The Academy of Management Review 21 (4), 986–1010.
Mulki, J.P., Jaramillo, F., Locander, W.B., 2006. Emotional exhaustion and organiza-
tional deviance: can the right job and a leader’s style make a difference? Journal
of Business Research 59 (12), 1222–1230.
Narver, J.C., Slater, S.F., 1990. The effect of a market orientation on business prof-
itability. Journal of Marketing 54 (4), 20–35.
Parameswaran, R., Yaprak, A., 1987. A cross-national comparison of consumer
research measures. Journal of International Business Studies 18 (1), 35–49.
Phillips, B., Tan, T.T.W., Julian, C., 2007. The theoretical underpinnings of emotional
dissonance: a framework and analysis of propositions. Journal of Service Mar-
keting 20 (7), 471–478.
Podsakoff, P.M., Organ, D.W., 1986. Self-reports in organizational research: problems
and prospects. Journal of Management 12 (4), 531–544.
Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y., Podsakoff, N.P., 2003. Common method
biases in behavioral research: a critical review of the literature and recom-
mended remedies. Journal of Applied Psychology 88 (5), 879–903.
Preacher, K.J., Hayes, A.F., 2004. SPSS and SAS procedures for estimating indirect
effects in simple mediation models. Behavior Research Methods, Instruments,
& Computers 36 (4), 717–731.
Pugh, D., 2001. Service with a smile: emotional contagion in service encounters.
Academy of Management Journal 44 (5), 1018–1027.
Rafaeli, A., Sutton, R.I., 1987. Expression of emotion as part of the work role. Academy
of Management Review 12 (1), 23–37.
Rodriguez, I., Bravo, M., Peiro, J., Schaufeli, W., 2001. The demands control support
model, locus of control and job dissatisfaction: a longitudinal study. Work &
Stress 15 (2), 97–114.
Ross, F., 1993. Type, severity and incidence of work stressors among Australian hos-
pitality industry employees. Australian Journal of Leisure and Recreation 3 (4),
5–12.
Rust, R.T., Zahorik, A.J., Keiningham, T.L., 1996. Service Marketing. Harper Collins,
New York, NY.
Singh, J., 2000. Performance productivity and quality of frontline employees in ser-
vice organizations. Journal of Marketing 64 (2), 15–34.
Spector, P.E., Chen, P.Y., O’Connell, B.J., 2000. A longitudinal study of relations
between job stressors and job strains while controlling for prior negative affec-
tivity and strains. Journal of Applied Psychology 85 (2), 211–218.
1082 G. Lee et al. / International Journal of Hospitality Management 31 (2012) 1067–1082
Staw, B.M., Bell, N.E., Clausen, J.A., 1986. The dispositional approach to job atti-
tudes: a lifetime longitudinal test. Administrative Science Quarterly 31 (1),
56–77.
Staw, B.M., Ross, J., 1985. Stability in the midst of change: a dispositional approach
to job attitudes. Journal of Applied Psychology 70 (3), 469–480.
Stone-Romero, E.F., 1994. Construct validity issues in organizational behavior
research. In: Greenberg, J. (Ed.), Organizational Behavior: The State of the Sci-
ence. Lawrence Erlbaum, Hillsdale, NJ, pp. 155–179.
Sutton, R.I., Rafaeli, A., 1988. Untangling the relationship between displayed emo-
tions and organizational sales: the case of convenience stores. Academy of
Management Journal 31 (3), 461–487.
Thoits, P.A., 1986. Social support as coping assistance. Journal of Consulting and
Clinical Psychology 54 (4), 416–423.
Thoresen, C.J., Kaplan, S.A., Barsky, A.P., Warren, C.R., De Chermont, K., 2003. The
affective underpinnings of job perceptions and attitudes: a meta-analytic review
and integration. Psychological Bulletin 129 (6), 914–945.
Totterdell, P., Holman, D., 2003. Emotion regulation in customer service roles: testing
a model of emotional labor. Journal of Occupational Health Psychology 8 (1),
55–73.
Van De Vliert, E., Van Yperen, N.C., 1996. Why cross-national differences in role over-
load? Don’t overlook ambient temperature! Academy of Management Journal
39 (4), 986–1004.
Watson, D., Clark, L.A., 1984. Negative affectivity: the disposition to experience
aversive emotional states. Psychological Bulletin 96 (3), 465–490.
Watson, D., Pennebaker, J., Folger, R., 1989. Beyond negative affectivity: measuring
stress and satisfaction in the workforce. Journal of Organizational Behaviour
Management 8 (2), 141–157.
Young, C.A., Corsun, D.L., 2009. What a nuisance: controlling of negative affectiv-
ity versus personality in hospitality stress research. International Journal of
Hospitality Management 28 (2), 280–288.
Zapf, D., Seifert, C., Schmutte, B., Mertini, H., Holz, M., 2001. Emotion work and
job stressors and their effects on burnout. Psychology and Health 16 (5),
527–545.
Zapf, D., 2002. Emotional work and psychological well-being: a review of the lit-
erature and some conceptual considerations. Human Resource Management
Review 12 (2), 237–268.
Zohar, D., 1994. Analysis of job stress profile in the hotel industry. International
Journal of Hospitality Management 13 (3), 219–231.