58
Journal of Marketing
Vol. 70 (July 2006), 58–73
© 2006, American Marketing Association
ISSN: 0022-2429 (print), 1547-7185 (electronic)
Thorsten Hennig-Thurau, Markus Groth, Michael Paul, & Dwayne D. Gremler
Are All Smiles Created Equal? How
Emotional Contagion and Emotional
Labor Affect Service Relationships
In this study, the authors examine the effects of two facets of employee emotions on customers’ assessments of
service encounters. Drawing on emotional contagion and emotional labor theories, they investigate the influence of
the extent of service employees’ display of positive emotions and the authenticity of their emotional labor display
on customers’ emotional states and, subsequently, on customers’ assessments of the service interaction and their
relationship with the service provider. To test the study hypotheses, 223 consumers participated in a simulated
service encounter in which actors played the roles of service employees. In a 2 × 2 factorial design, the employees
varied both the extent of their smiling behavior and their emotional labor display by engaging in surface or deep
acting. The results show that the authenticity of employees’ emotional labor display directly affects customers’
emotional states. However, contrary to expectations, the extent of employee smiling does not influence customer
emotions, providing no support for the existence of primitive emotional contagion in service interactions.
Furthermore, employee emotions exert an influence on customer outcomes that are of interest to marketers.
Thorsten Hennig-Thurau is Professor of Marketing and Media Research
(e-mail: tht@medien.uni-weimar.de), and Michael Paul is a doctoral stu-
dent (e-mail: michael.paul@medien.uni-weimar.de), Department of Mar-
keting and Media Research, College of Media, Bauhaus-University of
Weimar, Germany. Markus Groth is a senior lecturer, Department of Orga-
nizational Behavior, Australian Graduate School of Management, Univer-
sity of New South Wales (e-mail: markusg@agsm.edu.au). Dwayne D.
Gremler is Associate Professor of Marketing, Department of Marketing,
College of Business Administration, Bowling Green State University (e-
mail: gremler@bgsu.edu). The authors thank the students involved in the
data collection for their support, especially Johanna Pauge, Farnoush
Pourebrahimzadeh, and Vincent Charles for their acting performances.
They also thank the three anonymous JM reviewers, Alicia Grandey, Doug
Pugh, and Anna Mattila for their helpful and constructive comments on
previous versions of this article. They are grateful to CineStar and Con-
corde Home Entertainment for contributing the incentives for the partici-
pants in the study.
To read or contribute to reader and author dialogue on this article, visit
http://www.marketingpower.com/jmblog.
In general, the interaction between service employees andcustomers is considered an essential part of both cus-tomers’ assessments of service quality and their relation-
ship with the service provider (Bitner 1990; Gwinner,
Gremler, and Bitner 1998; Parasuraman, Zeithaml, and
Berry 1985). Despite the considerable amount of empirical
research on service relationships and customer assessments
of service quality, several aspects of the service interaction
have remained unexplored. An area of particular interest is
the role of emotions in service encounters. Although the
notion of having a friendly service staff and providing “ser-
vice with a smile” has become a generally unquestioned
mantra for service firms, empirical research about how
employees’ emotional states affect customers and their
assessments of service encounters has emerged only in
recent years.
Two research streams that address the role of emotions
in service encounters involve emotional contagion (Hat-
field, Cacioppo, and Rapson 1994) and emotional labor
(Hochschild 1983). “Emotional contagion” is defined as the
flow of emotions from one person to another, with the
receiver “catching” the emotions that the sender displays
(Schoenewolf 1990). In the context of service interactions,
emotional contagion creates a ripple effect of emotions
from service employees to customers (Pugh 2001; Tsai and
Huang 2002; Verbeke 1997). In other words, employees
who smile at customers may be contagious, in that they
change the affective state of customers and thus influence
customers’ perceptions and evaluations of the service
encounter. “Emotional labor” refers to service employees’
display of expected emotions as a self-regulatory process
(Hochschild 1983). When displaying expected emotions to
customers, employees can choose between two acting
strategies, surface or deep acting, which differ mainly in
their extent of authenticity (Grandey 2003).
The purposes of this study are twofold. First, we
attempt to extend marketing theory by developing and test-
ing a model of how employee emotions affect customers.
Building on emotional contagion and emotional labor
theories, we develop a model that enables us to test the dif-
ferential effects of two facets of employee emotions—
employee smiling behavior and authenticity of the emo-
tional labor display—on changes in customer affect in a
service setting. We also examine key customer conse-
quences of interest to service firms, including the impact of
employee emotions on customer satisfaction, customer–
employee rapport, and customers’ future loyalty intentions,
all of which constitute previously untested relationships.

How Emotional Contagion and Emotional Labor Affect Service Relationships / 59
A second contribution of this study is our use of an
experimental research design that enables us to assess the
cause–effect nature of emotional contagion processes more
precisely. The extant services literature on emotional conta-
gion has not answered the question whether customers
really catch employee emotions. Responding to Pugh’s
(2001, p. 1026) statement that “in the absence of a true
experimental design, it is not possible to conclude that con-
tagion alone is responsible [for the variation in customer
consequences],” our research establishes such causal rela-
tionships and fills an important gap in the customer–
employee emotions literature.
We first review the concepts that are central to our
study, namely, emotional contagion, emotional labor, and
customer consequences of service interactions, and then we
present a conceptual model and discuss the relationships
among the model elements. Next, we describe the design of
our experiment, which enables us to examine the relation-
ships of the conceptual model using a two-way repeated
measures analysis of variance (ANOVA) and partial least
squares (PLS) structural equation modeling. Finally, we
present the findings and their implications for services
research and management.
The Role of Emotions in Service
Interactions
Emotional Contagion
Research on emotional contagion attempts to explain how
emotions are transmitted among people in social interac-
tions and how “catching” another person’s emotions affects
the dynamics of the social interaction. Emotional contagion
can occur at both subconscious and conscious levels
(Barsade 2002). That is, the process of emotional contagion
can be attributed to people’s “tendency to automatically
mimic and synchronize facial expressions, vocalizations,
and movements with those of another person and, conse-
quently, to converge emotionally” (Hatfield, Cacioppo, and
Rapson 1994, p. 5) and to more conscious social compari-
son processes between people (Barsade 2002).
With “primitive emotional contagion” (Hatfield,
Cacioppo, and Rapson 1994), the transfer of emotions from
one person to another is the result of the receiver’s uncon-
scious, emotive processes. This type of emotional contagion
is driven by a two-step mimicry process, in which a person
(1) spontaneously imitates another person’s facial expres-
sions and other nonverbal cues, which (2) leads the person
to experience the corresponding emotions through physio-
logical links. Although the person feels the emotions that
result from mimicry, the processes that lead to this emotion
are often “subconscious and automatic” (Barsade 2002, p.
648). As a consequence, emotional contagion theories sug-
gest that primitive emotional contagion is spurred by the
extent to which the sender displays emotions; a greater
emotional display by the sender results in higher levels of
emotional contagion in the receiver.
In contrast, “conscious emotional contagion” is based
on social comparison processes in which people actively
search for emotions as a type of social information (Salan-
1This type of emotional contagion is sometimes referred to as
“emotional comparison” in the literature (e.g., Bartel and Saavedra
2000).
cik and Pfeffer 1978). This search activity is considered a
fundamental human behavior, which grows particularly
strong in situations perceived as ambiguous (Gump and
Kulik 1997). Specifically, conscious emotional contagion
theory argues that people compare their mood with another
person’s mood and adopt the sender’s emotive level when it
appears appropriate (Barsade 2002).1 For example, in the
absence of other social information, people visiting an
attorney for the first time can be expected to observe the
attorney’s emotional display and then to adopt his or her
emotions as a result of their desire to search for social infor-
mation and reduce perceived ambiguity. Unlike primitive
emotional contagion, conscious emotional contagion is
determined less by the extent to which the sender displays
emotions during an interaction (e.g., frequency of smiling)
and more by the authenticity with which the emotions are
displayed (e.g., genuineness of a smile). When the receiver
perceives the sender’s emotional display as fake or disin-
genuous, he or she will not interpret the emotional display
as adequate for reducing perceived ambiguity, so conscious
emotional contagion is less likely to occur.
Emotional Labor
The concept of emotional labor, which we consider a poten-
tial driver of customers’ emotional states and subsequent
assessments of service interactions, refers to the “effort,
planning, and control needed to express organizationally
desired emotions during interpersonal transactions” (Morris
and Feldman 1996, p. 987). Recent management literature
has considered emotional labor in an effort to better under-
stand how service organizations can manage employees’
positive displays to customers. Furthermore, it is linked to
the existence of either explicit or implicit organizational
display rules (Rafaeli and Sutton 1987) that define which
emotions employees are expected to display and which they
should suppress in the course of interacting with customers.
In general, service employees are expected to align their
displayed emotions with organizationally desired emotions
through their choice of emotional labor strategies
(Hochschild 1983). With regard to specific emotional labor
strategies, scholars have drawn on Hochschild’s (1983) dis-
tinction between surface acting and deep acting as the pri-
mary framework for service employees. In “surface acting,”
an employee tries to change only his or her outward behav-
ior to exhibit the required emotions. Thus, surface acting
refers to the act of displaying an emotion that is not felt and
could involve both suppression of felt emotions and faking
of unfelt emotions. For example, when dealing with an
angry or annoying customer, an employee may simply put
on a smile and pretend to be cheerful and friendly without
actually feeling the emotions. In other words, surface acting
constitutes the expression of feigned emotions and lacks
authenticity (Grandey 2003). With “deep acting,” however,
employees express expected (or required) emotions by
attempting to create these emotions within themselves. This

60 / Journal of Marketing, July 2006
strategy is similar to the method-acting technique developed
by Russian director Constantin Stanislavski (1965), in
which actors are taught to create self-induced true emotions
by using their emotional memory and recalling prior experi-
ences and emotions (Hochschild 1983). As an example of
deep acting in the context of service delivery, Hochschild
reports on flight attendants who are trained to deal with
angry passengers by thinking of them as frightened, first-
time fliers. This process enables the flight attendants to
change their inner feelings toward the customer from
annoyance to pity and sympathy. Thus, the distinction
between surface acting and deep acting is conceptually
aligned with the “service-as-theater” metaphor, which pos-
tulates the interaction between service employees and cus-
tomers as a dramaturgical interaction in which actors (i.e.,
employees) perform (i.e., provide service) on stage (i.e., the
service environment) in front of an audience (i.e., cus-
tomers) (Grove and Fisk 1992).
Consequences of Service Interactions
The display of employee emotions and the resultant cus-
tomer emotions likely affect various outcomes of interest to
marketing managers. We focus on three major customer
consequences of service interactions that are considered
particularly relevant to service companies: customer satis-
faction, customer–employee rapport, and customers’ future
loyalty intentions.
“Customer satisfaction” is widely regarded as the cogni-
tive assessment of a customer’s emotional experience (Hunt
1993). As Oliver (1981) discusses, satisfaction is consump-
tion specific; that is, it is related to a single consumption
experience. This transaction-related characteristic is often
considered the main difference between satisfaction and
similar evaluative concepts, such as consumer attitude and
perceived service quality, which are regularly modeled as
overall constructs or general evaluations of a service and are
unrelated to a specific consumption episode (Hennig-
Thurau and Klee 1997).
“Customer–employee rapport” is “a customer’s percep-
tion of having an enjoyable interaction with a service
provider employee, characterized by a personal connection
between the two interactants” (Gremler and Gwinner 2000,
p. 92). As a relational concept applicable to service settings,
rapport depends on one or more interactions between
employees and customers. Similar to satisfaction,
customer–employee rapport represents a customer’s cogni-
tive assessment of an affective state. It also is important to
stress that rapport can be cultivated through a single service
interaction and does not depend on a shared long-term his-
tory. Customer–employee rapport has been identified as a
salient issue for service organizations because rapport
exerts a strong influence on customer perceptions of service
delivery and service organizations (DeWitt and Brady 2003;
Gremler and Gwinner 2000).
Finally, customers’ “future loyalty intentions” constitute
a central component of service loyalty (Oliver 1997), which
itself is defined in various ways. Most definitions focus on a
customer’s willingness to visit a particular firm again
because of his or her positive emotions and cognitions
(Oliver 1999). In empirical research, future intentions repre-
sent a frequently studied component of loyalty, when loy-
alty is defined and/or measured as future intentions (e.g.,
Fornell 1992; Rust and Zahorik 1993; Zeithaml, Berry, and
Parasuraman 1996). We emphasize future loyalty intentions
because we believe that, in general, the future behavior of
customers is of more interest to service marketers than cur-
rent consumer attitudes (i.e., attitudinal loyalty) and/or prior
behavior.
A Conceptual Model of the Impact
of Employee Emotions on
Customers in Service Interactions
The impact of different facets of employees’ emotional dis-
plays on customer emotions and their consequences for cus-
tomers constitute the focus of this research. In Figure 1, we
show the conceptual model and the specific hypotheses
tested herein. In the following section, we discuss the pro-
posed relationships in detail.
Impact of Employee Emotions on Customer
Emotions
Drawing on emotional contagion and emotional labor
theories, we contend that in interpersonal interactions
between employees and customers, the likelihood that an
employee’s emotions affect customer emotions is facilitated
by two key variables: the extent of an employee’s smiling
and the authenticity of his or her emotional labor display.
Although prior research has found that employees who
deliver “service with a smile” increase a customer’s service
experience (e.g., Rafaeli and Sutton 1989; Tsai and Huang
2002), little is known about how or why an employee’s dis-
play of emotions is related to customer consequences. Emo-
tional contagion theory (Hatfield, Cacioppo, and Rapson
1994) suggests that people’s expression of positive emo-
tions facilitates a corresponding emotional state in others.
McHugo and colleagues (1985) demonstrate that exposure
to images of smiling faces produces corresponding
observed and self-reported emotions in study participants.
Furthermore, as we discussed previously, primitive (i.e.,
subconscious) emotional contagion suggests that the extent
of a service employee’s positive emotional display (e.g.,
amount of smiling) is the key driver of emotional contagion
(Hatfield, Cacioppo, and Rapson 1994). Thus, on the basis
of the assumption that customers perceive employee smil-
ing, we argue that if an employee increases his or her
amount of smiling, customers are more likely to mimic
these facial expressions subconsciously during the
encounter, thus altering their own emotional state. We use
the term “customer positive affect” for this positive emo-
tional state (DeWitt and Liu 2002; Watson and Tellegen
1985).
H1: High amounts of employee smiling lead to a greater
increase in customer positive affect than do low amounts
of employee smiling.
We also expect that the authenticity of the service
employee’s emotional labor display influences the cus-
tomer’s emotional state. As we indicated previously, deep
acting and surface acting are alternative strategies that

How Emotional Contagion and Emotional Labor Affect Service Relationships / 61
Customer–
employee
rapport
Customer
satisfaction with
transaction
Future loyalty
intentions
Authenticity
of the emotional
labor display
Extent of
employee
smiling
Change in
customer
positive
affect
H1
H2
H8
H7
H9
H10
H11
H4
H3
H5
H6
Employee Emotions
Customer ConsequencesCustomer Emotions
FIGURE 1
A Conceptual Model of Employee and Customer Emotions in Service Interactions
employees use to ensure that their emotional display toward
customers conforms to organizational display rules. A key
difference between these two emotional labor strategies is
the degree of authenticity of the emotional display (Broth-
eridge and Grandey 2002; Hochschild 1983; Kruml and
Geddes 2000). We argue that a high level of authenticity of
the employee’s emotional labor display, a main characteris-
tic of deep acting, triggers positive emotions within cus-
tomers due to their preference for being treated in an honest
and authentic way. In contrast, we do not expect such posi-
tive customer emotions when the employee’s emotional dis-
play is inauthentic, which is a main characteristic of surface
acting. Evidence from the social psychological literature on
authentic (i.e., Duchenne) smiles and emotion recognition
supports this view. For example, Ekman and colleagues
(Ekman 1992; Ekman, Davidson, and Friesen 1990; Ekman
and Friesen 1982) show that authentic smiles stimulate
more positive emotional reactions by respondents than do
“faked” smiles. It is argued that this distinction in reactions
is due to different neurological bases, in that observers
often respond more positively to some subtle facial cues
associated with authentic emotional displays, including the
symmetry of the smile or the activation of certain muscle
groups around the eyes (Ekman and Friesen 1982; Ekman,
Friesen, and O’Sullivan 1988). Thus, customer reactions to
authentic emotional displays are likely to be more positive
than those to inauthentic displays.
This argument receives additional support from con-
scious emotional contagion theory (Barsade 2002; Bartel
and Saavedra 2000), according to which the lack of authen-
ticity associated with employee surface acting makes it less
likely that customers can reduce their service-related ambi-
guity through the adoption of the employee’s emotions
(Grandey et al. 2005; Kruml and Geddes 2000). In the case
of deep acting, however, employees summon true and genu-
ine emotions from within and thus display positive emo-
tions to customers that are authentic, enabling those cus-
tomers to adopt the employee’s emotions consciously—that
is, to use the employee’s displayed emotions as a type of
social information to reduce the ambiguity associated with
the service experience. In other words, the change in cus-
tomer positive affect should be greater when employees’
emotional labor displays are authentic (i.e., engage in deep
acting) than when they are inauthentic (i.e., surface acting):

62 / Journal of Marketing, July 2006
H2: A high level of authenticity of the employee’s emotional
labor display (i.e., deep acting) leads to a greater increase
in customer positive affect than does a low level of
authenticity of the emotional labor display (i.e., surface
acting).
Customer Consequences of Employee Emotions
In addition to the proposed effects on customer positive
affect, we also expect that the independent variables (i.e.,
extent of employee smiling and authenticity of the emo-
tional labor display) directly affect both customer satisfac-
tion with the transaction and customer–employee rapport.
Regarding customer satisfaction, customers often interpret
an employee’s emotional display as part of the service itself
(i.e., the notion of service as theater; Grove and Fisk 1992),
which suggests that they hold expectations about the dis-
play of positive emotions (i.e., smiling) that influence their
level of satisfaction (Tsai 2001). Regarding customer–
employee rapport, an employee’s smiling may be an
antecedent of rapport, in that it increases the receiver’s
enjoyment of the personal interaction (Gillis, Bernieri, and
Wooten 1995; Tickle-Degnen and Rosenthal 1990).
H3: High amounts of employee smiling lead to a greater
increase in customer satisfaction with the transaction than
do low amounts of employee smiling.
H4: High amounts of employee smiling lead to a greater
increase in customer–employee rapport than do low
amounts of employee smiling.
Similarly, we expect that the provision of authentic
emotions as part of employee deep acting leads to greater
customer satisfaction and rapport than does an inauthentic
emotional labor display (i.e., surface acting). Grandey
(2003) finds that employees’ use of deep acting leads to
higher ratings of service delivery than does the use of sur-
face acting, and Grandey and colleagues (2005) report that
customer satisfaction is higher when customers perceive
employee behavior as authentic. Grandey (2003) also pro-
vides support for the direct impact of the type of emotional
labor strategy on customer-perceived rapport. In her study,
employees’ deep acting is related to perceptions of friendli-
ness and warmth, both of which are considered characteris-
tics of rapport (Gremler and Gwinner 2000).
H5: A high level of authenticity of the employee’s emotional
labor display (i.e., deep acting) leads to a greater increase
in customer satisfaction with the transaction than does a
low level of authenticity of the employee’s emotional
labor display (i.e., surface acting).
H6: A high level of authenticity of the employee’s emotional
labor display (i.e., deep acting) leads to a greater increase
in customer–employee rapport than does a low level of
authenticity of the employee’s emotional labor display
(i.e., surface acting).
Relationships Between Customer Emotions and
Customer Consequences
Theoretical support for the impact of a customer’s emo-
tional state on his or her satisfaction comes from service
research, which suggests that customer satisfaction with a
service encounter is strongly influenced by customer emo-
tions (Oliver 1997). In particular, when customers assess a
specific consumption experience (i.e., form their level of
satisfaction with the service), they draw strongly on their
current emotional state and ask themselves questions such
as “How do I feel about it?” Thus, a change in a customer’s
emotions due to an employee’s emotional display should
influence the customer’s satisfaction. This impact is consis-
tent with the affect-as-information model from social psy-
chology (Schwartz and Clore 1988), which suggests that
people rely on their moods as information cues when they
make evaluative judgments. Consequently, a change in cus-
tomer emotions facilitated by an employee’s emotional dis-
play should lead to a change in customer satisfaction, such
that an increase in customer positive affect results in higher
levels of satisfaction.
H7: An increase in customer positive affect has a positive
effect on customer satisfaction with the transaction.
We further hypothesize that an increase in customer
positive affect as a result of employee emotions will influ-
ence customers’ perceptions of their rapport with the
employee. Specifically, when a customer is “infected” by an
employee’s positive emotions as a result of a service inter-
action, the customer likely will enjoy the interaction with
the service employee to a greater extent. Because an enjoy-
able interaction is a key characteristic of customer–
employee rapport (Gremler and Gwinner 2000), an increase
in customers’ positive emotions should ultimately lead to
higher levels of rapport.
H8: An increase in customer positive affect has a positive
effect on customer–employee rapport.
The services literature also indicates relationships
among customer–employee rapport, customer satisfaction
with the transaction, and future loyalty intentions. Specifi-
cally, we argue that rapport is positively related to customer
satisfaction because an enjoyable interaction with a high
degree of customer–employee rapport is usually one in
which customers reveal personal information, which
enables employees to customize the service offering to the
customer’s needs (Gremler and Gwinner 2000); both of
these elements are considered integral parts of customer sat-
isfaction. Furthermore, we expect that customer satisfaction
is positively related to customers’ future loyalty intentions
through the creation of positive attitudes toward the service
provider (Yi 1990), a claim we base on both attitude theory
(Fishbein and Ajzen 1975) and exit-voice theory
(Hirschman 1970). Finally, because customers who person-
ally like a service employee and have rapport with him or
her can be expected to form positive expectations about a
future service experience with this employee (Gremler and
Gwinner 2000), we propose that customer–employee rap-
port has a positive relationship to customers’ future loyalty
intentions.
H9: Customer–employee rapport is positively related to cus-
tomer satisfaction with the transaction.
H10: Customer–employee rapport is positively related to cus-
tomers’ future loyalty intentions.
H11: Customer satisfaction with the transaction is positively
related to customers’ future loyalty intentions.

How Emotional Contagion and Emotional Labor Affect Service Relationships / 63
Methodology
Participants
Participants in the experiment were undergraduate and
graduate students from various non–business schools of a
medium-sized university. The final sample contained data
from 223 participants, 46.8% of whom were women. The
mean age of the respondents was 23.5 years (SD = 3.2);
ages range from 18 to 48 years. Although the use of student
samples is sometimes considered a limitation in marketing
research, drawing on student samples in experimental
designs, especially those involving role playing, is well
accepted for examining causal relationships (e.g., Barsade
2002; Surprenant and Solomon 1987).
We contacted participants through the university’s Web
site and course announcements; they received either a free
movie ticket for a local movie theater or a free DVD movie
in return for their participation. Before the actual data col-
lection, we conducted a pretest with 32 consumers to assess
the adequacy of the study design and the experimental
manipulations and to ensure that the psychometric proper-
ties of our study measures were adequate.
Procedure
General research design. We informed all participants
that a new business model for a personalized movie rental
service was being tested at the university. The service was
positioned as a new “movie consulting service” for which
customers would answer a series of questions about their
movie-viewing habits and preferences in a one-to-one con-
sultation with a service employee (unbeknownst to the sub-
ject, this employee was a trained actor). Based on their
answers, customers would then receive personalized advice
for potential movie rentals from the employee. Throughout
the duration of the study, professionally designed posters
that promoted this new service appeared in various loca-
tions at the university. The university also set aside specific
facilities to serve as the movie consulting service location
for the duration of this study. The layout and interior deco-
ration of these facilities resembled those of actual video
rental services (e.g., a service counter, shelves containing
movies, movie posters, and life-sized figures announcing
current and upcoming movies). Donations of actual DVD
and VHS movies and various poster displays from local
video rental stores and movie theaters decorated the facili-
ties and thus increased the realism of the servicescape (see
Figure 2).
Experimental manipulations. Using a 2 × 2 between-
subjects factorial design, we manipulated the extent of
employee smiling (high versus low) and the authenticity of
the emotional labor display (high authenticity/deep acting
versus low authenticity/surface acting). We randomly
assigned participants to one of the four experimental condi-
tions, high smiling/deep acting (n = 58), high smiling/
surface acting (n = 52), low smiling/deep acting (n = 55),
and low smiling/surface acting (n = 58).
Dramaturgy. We used appointments to schedule one
participant at a time, which limited waiting time and elimi-
nated potential interactions among the participants. On
arrival, each participant filled out a preencounter question-
naire and then immediately entered the video consulting
store to begin the service encounter. During the service
encounter, participants interacted with the service
employee, who engaged them in a conversation by asking a
series of specific questions about movie preferences and
viewing habits. The service encounter concluded with the
employee recommending a specific movie on DVD or VHS,
which he or she then handed to the customer. In general,
these service encounters lasted between 5 and 10 minutes
(M = 7.5 minutes).
Actors. The service employees were three trained stu-
dent actors (two female, one male) whom we specifically
recruited for this study and paid for their time. We hired
these actors on the basis of their experience in service work,
their interest in acting, and their knowledge of motion pic-
tures. Initially, we narrowed the pool of potential candidates
down to six people, all of whom then performed in a series
of auditions and rehearsals and participated in the pretest of
the experiment. On the basis of their performances in these
tests, we selected the final three actors and trained them
over a period of six weeks.
Furthermore, we developed a role description for each
of the four conditions. All roles required the actors to
behave in a customer-oriented manner and avoid the display
of negative emotions. For the extent of smiling manipula-
tion, we trained the employees to smile frequently when
displaying a high level of smiling and to minimize their
smiling when displaying low-level smiling behavior. In line
with previous research, we defined smiling as a noticeable
upward twist of the employee’s lips (Pugh 2001; Sutton and
Rafaeli 1988). For the authenticity of the emotional labor
display manipulation, we trained the actors in accordance
with the characteristics of surface and deep acting, as
described in the literature. They read some pertinent intro-
ductory literature (e.g., Grandey 2003; Hochschild 1983),
and two of the authors then discussed the concepts with the
actors for a total of ten hours. Mainly, we drew from the
concept of method acting by showing movie excerpts and
using teaching techniques associated with method acting.
For the surface acting role (i.e., low authenticity of the emo-
tional labor display), the employee was instructed to adapt
only his or her outward behavior to the customer’s needs
but not his or her inner feelings. For deep acting (i.e., high
authenticity of the emotional labor display), the employee
was instructed to create the appropriate emotions within
him- or herself. To teach these actors the skills to do so
effectively, we employed specific exercises used in
Stanislavski’s (1965) acting technique, which are com-
monly taught to acting students as part of their basic train-
ing. These exercises target the development of specific
skills for using emotion memories (i.e., remembering and
reexperiencing specific emotions) to evoke the required
emotions. We continued the training until both the actors
and the researchers were satisfied that the actors had mas-
tered a sufficient skill set to enable them to act out the spe-
cific roles required for the experiment.

64 / Journal of Marketing, July 2006
FIGURE 2
Design of the Video Consulting Store
Finally, we developed a standardized service script that
consisted of a list of questions about customers’ movie-
related preferences to ensure that all service encounters
would be identical in nature in all respects other than the
experimental manipulations (for further details about the
service script, see Appendix A). Data collection took place
over a period of seven days. Each actor played only one role
per day over a time of approximately two-and-a-half hours.
We randomly determined the roles in advance, and each
actor performed for roughly equal amounts of time in each
of the four roles.
Measures
Preencounter questionnaire. Study participants received
a questionnaire just before they entered the video rental
store, which included four items from Brief and colleagues’
(1988) job affect scale to measure preencounter customer
positive affect (see also Burke et al. 1989). We dropped two
of the original six items (strong and active) because we
deemed them to be inappropriate for use by customer
respondents. In addition, to conceal the true nature of the
study, the preencounter questionnaire contained various
2There is an ongoing discussion about the problems associated
with the use of difference scores (e.g., Peter, Churchill, and Brown
1993). These issues predominantly address the integration of two
measures (e.g., service expectations and service performance
evaluation) but are less relevant when the same construct is mea-
sured at two points in time. Two common concerns about differ-
ence scores are that they lack variability and do not differ from the
initial level (Tisak and Smith 1994). However, these issues do not
filler items about preferences and consumption behaviors
for motion pictures.
Postencounter questionnaire. Participants completed a
postencounter questionnaire immediately after the end of
the service encounter; it included measures of post-
encounter customer positive affect, customer–employee
rapport, customer satisfaction with the transaction, and
future loyalty intentions, as well as demographic variables.
We assessed postencounter customer positive affect with
the same items used in the preencounter survey and calcu-
lated any changes as postencounter customer positive affect
less preencounter customer positive affect for each of the
four items.2 To measure rapport, we used four items from

How Emotional Contagion and Emotional Labor Affect Service Relationships / 65
the scale developed by Gremler and Gwinner (2000); we
excluded items that implicitly suggest a long-term relation-
ship between the customer and the employee. For customer
satisfaction with the transaction, we used five items from
the work of Churchill and Surprenant (1982) and Bitner and
Hubbert (1994). Finally, we measured future loyalty inten-
tions with four items, three of which we adapted from the
work of Zeithaml, Berry, and Parasuraman (1996) and one
from the work of Taylor and Baker (1994). We measured all
items using seven-point scales, in which higher numbers
indicated greater agreement with the construct (for a list of
all study items, see Appendix B).
Results
Manipulation Checks
We employed several methods to ensure the validity of our
study manipulations. First, unbeknownst to customers, all
service encounters were recorded with a hidden digital
video camera and transmitted in real time to an adjacent
room. Two trained judges (one of the authors and one grad-
uate student) watched each interaction to judge whether it
was consistent with the required experimental manipula-
tion. Service encounters that the judges deemed to be incon-
sistent with the required role were dropped from further
analyses. Second, because the authenticity of the emotional
labor display manipulation ultimately depends on an
employee’s self-evaluation, after each service encounter, we
asked the service employees whether they believed they had
effectively portrayed the desired role. In addition, during
the debriefing, one participant guessed the true nature of the
study. As a result of these measures, we excluded three ser-
vice encounters from further analysis, which resulted in 223
usable encounters.
Third, after completion of the study, four graduate stu-
dents (two male, two female) who were not familiar with
the study watched all 223 videotaped service encounters
and rated each interaction on several dimensions.3 Specifi-
cally, two of the judges separately watched all interactions
and recorded (1) the frequency of employee smiles (defined
as the number of times the employee noticeably smiled)
during each interaction and (2) the duration of each smile
(measured in seconds). We transformed these data into a
smiles-per-minute measure and a measure of the percentage
of time the employee smiled during each encounter, which
enabled us to control for service encounter length. The
interrater agreement, which we measured with the intraclass
correlation coefficient (ICC; Shrout and Fleiss 1979), was
high between the two judges for both items (frequency of
smiles: ICC = .909, p < .01; duration of smiles: ICC = .948,
p < .01). The ANOVA results also showed that the manipu-
lation of the extent of employee smiling had a significant
effect on the frequency of smiles per minute (Mhigh extent =
3.5, Mlow extent = .4; F(1, 220) = 386.62, p < .01) and the
percentage of time spent smiling (Mhigh extent = 35.2%,
Mlow extent = 1.8%; F(1, 220) = 417.09, p < .01), as
intended. We found no significant effect for the authenticity
of the emotional labor display on the frequency of smiles
(Mhigh authenticity = 2.0, Mlow authenticity = 1.8; F(1, 220) =
1.08, not significant [n.s.]), though it influenced the per-
centage of time spent smiling (Mhigh authenticity = 23.9%,
Mlow authenticity = 12.6%; F(1, 220) = 18.64, p < .01), which
suggests a potential confounding of the manipulations. As
Perdue and Summers (1986, p. 323) suggest, providing evi-
dence that the effect sizes for unintended manipulations are
substantially smaller than those for the intended manipula-
tions indicates that the experimental manipulations have
worked effectively and that statistical significance “should
not be of great concern.” In our case, the extent of the
employee smiling manipulation had a much greater effect
on the percentage of time spent smiling (partial η2 = .656)
than did the authenticity of the emotional labor display
manipulation (partial η2 = .078), thus providing support for
the validity of our experimental manipulation.
The other two judges also watched all 223 videotaped
interactions and separately provided a global rating of the
authenticity of the employee’s emotional display for each
interaction (on a four-point scale ranging from “not at all
authentic” [1] to “very authentic” [4]). Again, interrater
agreement was high between the two judges (ICC = .927,
p < .01). The authenticity of the emotional labor display
manipulation had a significant effect on authenticity ratings
(Mhigh authenticity = 3.8, Mlow authenticity = 1.2; F(1, 220) =
2552.78, p < .01), as intended. The extent of employee
smiling also had a significant main effect on authenticity
ratings (Mhigh extent = 2.7, Mlow extent = 2.3; F(1, 220) = 4.21,
p < .05), but the effect size of the authenticity of the emo-
tional labor display manipulation (partial η2 = .932) was
greater than that of the employee smiling manipulation
(partial η2 = .147), again providing support for the validity
of our experimental manipulation.
Reliability and Validity Assessment
We report the means, standard deviations, and correlation
coefficients of all variables in Table 1. To assess the relia-
bility and validity of our measures, we calculated Cron-
bach’s alpha coefficients for each construct. All alpha
scores are satisfactory, with no values below .80. In addi-
tion, we conducted a confirmatory factor analysis with all
multi-item measures in the model. The overall fit statistics
for the four-factor model indicate that the model provides
an acceptable fit to the data: χ2(129, N = 223) = 333.7, p <
.01; comparative fit index = .950; incremental fit index =
.950; Tucker–Lewis index = .941; standardized root mean
square residual = .048; and root mean square error of
approximation = .086.
The average variance extracted is greater than .60 for all
constructs, and the composite reliability measures are all
greater than .80. We also find support for convergent
validity because the t-values for all constructs are signifi-
cant at p < .01 (Anderson and Gerbing 1988). Similarly, we
appear to be critical in our case. Specifically, the standard devia-
tions of the difference scores range from 1.46 to 1.65, and the cor-
relation between the difference score and preencounter affect
(–.34), though significant, is smaller than 1.
3We thank the anonymous reviewers for their suggestions of
how to use the videotaped encounters to provide additional
manipulation checks for this study.

66 / Journal of Marketing, July 2006
TABLE 1
Descriptive Statistics, Reliability Information, and Correlations
Number
of Items M SD 1 2 3 4 5 6
1. Customer positive affect (preencounter) 4 3.70 1.20 .812
2. Customer positive affect (postencounter) 4 4.41 1.41 .556 .904
3. Change in customer positive affect 4 .71 1.24 –.336 .596 .806
4. Customer–employee rapport 4 4.91 1.66 .233 .643 .503 .967
5. Customer satisfaction with transaction 5 4.69 1.47 .323 .771 .561 .827 .932
6. Future loyalty intentions 4 5.12 1.56 .269 .662 .486 .696 .798 .940
Notes: N = 223. Values along the diagonal represent Cronbach’s alpha internal consistency estimates. All correlations are significant at p < .01
(two tailed).
4Although we did not formally hypothesize an interaction
effect, our examination indicates that both effects are largely inde-
pendent with no interaction effect (F(1, 217) = .607, n.s.). We also
investigate the extent to which the results are influenced by the
individual employees by controlling for actors in the analysis. This
did not change the results.
find support for discriminant validity; the squared correla-
tions between each pair of constructs are smaller than the
average variance explained of the respective constructs
(Fornell and Larcker 1981). Thus, these results indicate the
acceptable reliability, convergent validity, and discriminant
validity of our measures.
Two-Way Repeated Measures ANOVA
To test H1 and H2, we conducted a two-way repeated mea-
sures ANOVA to examine the effects of the extent of
employee smiling and the authenticity of the emotional
labor display on the change in customer positive affect. The
within-subjects factor, customer positive affect, contains
two levels: preencounter affect (Time 1) and postencounter
affect (Time 2). As Tables 2 and 3 show, the extent of
employee smiling has no significant main effect on cus-
tomer positive affect (F(1, 217) = 1.5, n.s.), but the authen-
ticity of the emotional labor display by employees has a sig-
nificant effect (F(1, 217) = 35.2, p < .01, partial η2 = .139).4
The nature of this effect is in the expected direction:
Employees who express authentic emotions by engaging in
deep acting facilitate customer positive affect to a much
greater extent (change in customer positive affect = 1.18,
SD = 1.09) than do those who display inauthentic emotional
labor by engaging in surface acting (change in customer
positive affect = .25, SD = 1.22). This finding holds true not
only for the overall composite score of customer affect but
also for each of the four emotions (i.e., elated, peppy, enthu-
siastic, and excited). Thus, we find strong evidence in sup-
port of H2 but not of H1. Therefore, we conclude that
employees’ emotional displays affect customers’ emotional
states, but this process appears to be driven primarily by the
authenticity of the emotional labor display rather than by
primitive emotional contagion through employee smiling.
PLS Equation Modeling
Next, we tested all elements of our conceptual model simul-
taneously with PLS structural equation modeling. As a
component-based method, PLS permits the use of nominal
data, which we need to assess the effects of smiling and
5Significance levels and the direction of the effects for all
hypothesized paths are identical to those of the difference score
model. The variance explanations for the postencounter model are
11.6% for postencounter customer positive affect, 60.4% for rap-
port, 78.4% for customer satisfaction, and 66.0% for loyalty
intentions.
6We generated the t-values through a bootstrapping procedure
with 223 resamples with 100 cases each (Fornell and Bookstein
1982).
authenticity (Fornell and Bookstein 1982). Furthermore, as
a distribution-free method, PLS has fewer constraints and
statistical specifications than covariance-based techniques,
such as LISREL.
We operationalized both the extent of employee smiling
and the authenticity of the emotional labor display as
dichotomous variables (1 = low extent, 2 = high extent; 1 =
low authenticity/surface acting, 2 = high authenticity/deep
acting, respectively). To estimate the model paths, we used
PLS Graph 3.0 and estimated the inner weightings with the
path method (Chin 2001). To estimate the predictive power
of the model, we applied a blindfolding approach (Fornell
and Bookstein 1982). This approach results in Q2 values of
.38 (change in customer positive affect), .67 (customer sat-
isfaction), .81 (rapport), and .72 (loyalty intentions), all of
which are significantly different from 0 and therefore indi-
cate that the model has predictive power (Geisser 1974;
Stone 1974). The model explains 14.8% of the variance in
change in customer positive affect, 47.6% of the variance in
rapport, 71.7% of the variance in customer satisfaction with
the transaction, and 65.9% of the variance in future loyalty
intentions, in further support of its relevance (Chin 1998).
All indicator reliabilities are greater than .75, composite
reliability is greater than .85 for all constructs, and the aver-
age variance explained is greater than .60 in all cases (for
more detailed information, see Appendix B). To determine
whether the use of a difference score measure for change in
positive affect influenced the results, we also analyzed an
alternative model with postencounter customer positive
affect rather than change in customer positive affect, but we
found no differences between the two models.5
The PLS results support most of our hypotheses about
the inner model relationships and provide convergent
validity for the findings of the two-way repeated measures
ANOVA (see Table 4).6 Specifically, our results show that
the authenticity of the employee’s emotional labor display

How Emotional Contagion and Emotional Labor Affect Service Relationships / 67
TA
B
LE
2
R
es
ul
ts
o
f t
he
T
w
o
-W
ay

R
ep
ea
te
d
M
ea
su
re
s
AN
OV
A
O
ve
ra
ll
Cu
st
om
er
Po
si
tiv
e
Af
fe
ct

El
at
ed
Pe
pp
y
En
th
us
ia
st
ic
Ex
ci
te
d
In
de
pe
nd
en
t V
ar
ia
bl
e
d.
f.
F
Pa
rt
ia
l ηη
2
F
Pa
rt
ia
l ηη
2
F
Pa
rt
ia
l ηη
2
F
Pa
rt
ia
l ηη
2
F
Pa
rt
ia
l ηη
2
Ex
te
nt
o
f e
m
pl
oy
e
e
s
m
ilin
g
(S
MI
LE
)
1
1.
54
9
2.
70
8
1.
30
9
.
19
0
1.
17
9
Au
th
en
tic
ity
o
f t
he
e
m
ot
io
na
l l
ab
or
di
sp
la
y
(A
UT
H)
2
35
.1
75
*
.
13
9
23
.5
65
*
.
09
8
12
.9
43
*
.
05
7
16
.1
85
*
.
07
0
31
.3
88
*
.
12
8
SM
IL
E
×
AU
TH
2
.
60
7
.
34
8
.
93
8
.
00
2
1.
04
9
Er
ro
r
21
7
(.6
66
)
(1.
19
3)
(1.
09
5)
(1.
01
0)
(1.
18
1)
*
p
<
.
01
.
N
ot
es
:W
e

ca
lc
ul
at
e
th
e
ov
e
ra
ll
co
m
po
sit
e
sc
or
e
fo
r
cu
st
om
er
p
os
itiv
e
a
ffe
ct
a
s
th
e
m
ea
n
va
lu
e
of
th
e
fo
u
r
e
m
o
tio
n
ite
m
s.
Va
lu
es
in
p
ar
en
th
es
es
a
re
m
ea
n
sq
ua
re
te
rm
s.

68 / Journal of Marketing, July 2006
TABLE 3
Means and Standard Deviations for Customer Positive Affect
Customer Positive Affect
Change in
Time 1: Time 2: Customer
Preencounter Postencounter Positive Affect
Extent of employee smiling High extent 3.72 (1.21) 4.54a (1.38) 0.82b (1.26)
Low extent 3.69 (1.20) 4.30a (1.44) 0.61b (1.21)
Authenticity of the emotional High authenticity 3.71 (1.20) 4.87a (1.28) 1.18b (1.09)
labor display Low authenticity 3.70 (1.21) 3.95a (1.40) 0.25b (1.22)
Notes: Standard deviations are shown in parentheses. Values with the same superscript are significantly different from each other at p < .001;
all other values do not significantly differ.
has a strong and significant impact on the change in cus-
tomer positive affect—high authenticity through deep act-
ing results in a significantly greater increase in the change
in customer positive affect than does low authenticity
through surface acting—but the extent of employee smiling
does not significantly influence the change in customer
positive affect.
The extent of employees’ smiling and the authenticity of
their emotional labor display both have a significant, direct
impact on customer–employee rapport, in support of H4 and
H6, but they do not affect customer satisfaction, which fails
to support H3 and H5. The PLS results also show that
change in customer positive affect has a significant effect
on both customer satisfaction with the transaction and
customer–employee rapport, in support of H7 and H8,
respectively. Rapport positively influences customer satis-
faction (in support of H9), and customer satisfaction posi-
tively influences future loyalty intentions (in support of
H11). Contrary to our expectations, rapport has no direct
relationship to future loyalty intentions, leading us to reject
H10. However, the total effect of rapport on loyalty inten-
tions is strong and significant as a result of the strong
impact of rapport on satisfaction (for a listing of all total
effects, see Table 4).
Discussion
What We Have Learned
The purpose of this study was to develop and experimen-
tally test a conceptual model of how employee emotions
influence customers in service encounters. We apply a two-
way repeated measures ANOVA and PLS structural equa-
tion modeling to data collected from a sample of 223 con-
sumers. The results of this research provide several
important contributions to the literature.
First, we fill an important gap in the service marketing
literature by providing empirical evidence that an
employee’s emotional display can trigger changes in a cus-
tomer’s affective state. Our study is the first to provide a
direct test of emotional contagion and its related processes
in service interactions, and the controlled experimental
design of our study enables us to test causal effects between
different facets of an employee’s display of emotions and
the customer’s state.
Second, our findings suggest that the authenticity of the
employee’s emotional labor display, rather than the extent
of smiling, influences the customer’s emotions and percep-
tions. Respondents in our study who encountered authentic
employees (i.e., those who engaged in deep acting) were far
more likely to adopt the emotions of that employee than
were those who interacted with inauthentic employees (i.e.,
those who engaged in surface acting). This finding con-
tributes to the emotional labor literature in that our study is
the first to provide evidence that different types of emo-
tional labor (i.e., deep versus surface acting) differentially
influence the customer experience during service encoun-
ters. Furthermore, our findings contribute to the emotional
contagion literature by contradicting the dominant belief
that emotional contagion in service settings is based mainly
on mimicry effects (i.e., primitive contagion processes; see,
e.g., Pugh 2001). Although mimicry effects occur within
extremely short time frames (often less than one second)
and thus might affect the receiver’s emotions for a short
period, our results suggest that their impact fades during the
course of a service encounter. That is, primitive emotional
contagion might occur in the early phases of service
encounters, but it does not remain throughout the comple-
tion of the encounter, because postencounter emotions do
not appear to be influenced by mimicry effects.
Third, our findings provide support for the affect-as-
information theory, which links customer emotions to satis-
faction (Schwartz and Clore 1988); the results show that
both the customer’s postencounter emotional state and
change in positive affect lead to increased customer satis-
faction. An increase in customer positive affect also influ-
ences customer satisfaction indirectly through customer–
employee rapport. Because satisfaction is positively related
to customers’ future loyalty intentions, employees’ emo-
tional displays appear to play important roles in a service
firm’s long-term success and have significant impacts on
key customer outcomes.
Fourth, we could not replicate Grandey and colleagues’
(2005) results of a direct relationship between emotional
labor and customer satisfaction. In our study, the two paths
from the extent of employee smiling and the authenticity of
the emotional labor display to customer satisfaction are not
significant. Therefore, our results suggest that such effects
are of an indirect nature, with customer–employee rapport
being the main driver of customer satisfaction. Thus, the

How Emotional Contagion and Emotional Labor Affect Service Relationships / 69
TA
B
LE
4
Pa
th
C
oe
ffi
ci
en
ts
fr
o
m
P
LS
A
na
ly
si
s
Pa
th

To
ta
l
H
yp
ot
he
si
s
H
yp
ot
he
si
s
Im
pa
ct
o
f
O
n
Co
ef
fic
ie
nt
t-V
al
ue
Im
pa
ct
Su
pp
or
te
d?
H
1
Ex
te
nt
o
f e
m
pl
oy
e
e
s
m
ilin
g
Ch
an
ge
in
c
us
to
m
er
p
os
itiv
e
a
ffe
ct
.
07
0
1.
08
1
.
07
0
N
o
H
2
Au
th
en
tic
ity
o
f t
he
e
m
ot
io
na
l l
ab
or
d
isp
la
y
Ch
an
ge
in
c
us
to
m
er
p
os
itiv
e
a
ffe
ct
.
37
5*
4.
57
2
.
37
5
Ye
s
H
3
Ex
te
nt
o
f e
m
pl
oy
e
e
s
m
ilin
g
Cu
st
om
er
s
at
isf
a
ct
io
n
w
ith
tr
a
n
sa
ct
io
n
–
.
04
6
–
1.
08
1
.
01
4
N
o
H
4
Ex
te
nt
o
f e
m
pl
oy
e
e
s
m
ilin
g
Cu
st
om
er
–e
m
pl
oy
e
e
r
a
pp
or
t
.
19
4*
2.
75
2
.
21
7
Ye
s
H
5
Au
th
en
tic
ity
o
f t
he
e
m
ot
io
na
l l
ab
or
d
isp
la
y
Cu
st
om
er
s
at
isf
a
ct
io
n
w
ith
tr
a
n
sa
ct
io
n
–
.
05
5
–
1.
10
0
.
38
4
N
o
H
6
Au
th
en
tic
ity
o
f t
he
e
m
ot
io
na
l l
ab
or
d
isp
la
y
Cu
st
om
er
–e
m
pl
oy
e
e
r
a
pp
or
t
.
45
3*
6.
65
4
.
57
4
Ye
s
H
7
Ch
an
ge
in
c
us
to
m
er
p
os
itiv
e
a
ffe
ct
Cu
st
om
er
s
at
isf
a
ct
io
n
w
ith
tr
a
n
sa
ct
io
n
.
20
6*
3.
04
1
.
45
3
Ye
s
H
8
Ch
an
ge
in
c
us
to
m
er
p
os
itiv
e
a
ffe
ct
Cu
st
om
er
–e
m
pl
oy
e
e
r
a
pp
or
t
.
32
3*
5.
04
1
.
32
3
Ye
s
H
9
Cu
st
om
er
–e
m
pl
oy
e
e
r
a
pp
or
t
Cu
st
om
er
s
at
isf
a
ct
io
n
w
ith
tr
a
n
sa
ct
io
n
.
76
5*
11
.3
80
.
76
5
Ye
s
H
10
Cu
st
om
er
–e
m
pl
oy
e
e
r
a
pp
or
t
Fu
tu
re
lo
ya
lty
in
te
nt
io
ns
.
10
9
1.
34
2
.
65
1
N
o
H
11
Cu
st
om
er
s
at
isf
a
ct
io
n
w
ith
tr
a
n
sa
ct
io
n
Fu
tu
re
lo
ya
lty
in
te
nt
io
ns
.
72
0*
7.
49
1
.
72
0
Ye
s
*
p
<
.
05
.

70 / Journal of Marketing, July 2006
link between authenticity of the emotional labor display and
customer satisfaction is not as direct as previously thought
but instead is mediated by the extent to which customers
perceive that rapport with the service employee exists. A
possible explanation of this finding is that whereas the con-
struct of rapport focuses mainly on the interaction with the
employee, satisfaction also captures the customer’s assess-
ment of other aspects of the service encounter, such as the
design of the servicescape (i.e., “I like how the video store
looks”) and outcome qualities (i.e., “I like the movie I was
given by the employee”).
Fifth, although we find that customer–employee rapport
is a driver of satisfaction, our results also suggest that rap-
port does not exert a direct influence on customer loyalty.
Although previous marketing research has postulated such a
direct relationship (Gremler and Gwinner 2000), our study
suggests that the interplay among rapport, satisfaction, and
future loyalty intentions may be more complex than previ-
ously believed.
What We Still Need to Learn
Although our findings expand the extant knowledge on
emotions in service encounters, we recognize several limi-
tations that must be taken into account when generalizing
our results. The design of our experiment provides insight
into the transfer of employee emotions in one specific ser-
vice domain—an innovative video rental service—but it
remains unclear whether findings would be similar for other
services. Another limitation is that we focus on positive
emotions because they are most relevant (and desired) in
the service delivery context. However, similar ripple effects
may occur for negative emotions. For example, anger and
frustration displayed by employees may transfer to cus-
tomers and negatively affect their service experience. Fur-
ther research should address this issue by testing whether
effects similar to those we find apply equally to the flow of
negative emotions in the service delivery context.
Because participants in our study engaged in a service
that was unfamiliar to them, their emotional processes were
completely independent of any prior service interactions.
However, in many real-life services, relationships between
employees and customers have been established over a
series of encounters, which might influence the emotional
flow in future service transactions. Another avenue for fur-
ther research would be to test the cross-cultural stability of
our findings. Given that cultural norms about service deliv-
ery and smiling behavior in general vary across cultures,
additional research should examine the extent to which cul-
tural variables influence the interplay of employee emo-
tional displays and customer variables in service deliveries.
What Marketing Managers Should Do
This study illustrates that the emotions displayed by front-
line employees are an important driver of the relationships
between service employees and customers. The transfer of
positive emotions from employees to customers can enable
firms to establish high levels of customer–employee rap-
port, customer satisfaction, and future loyalty intentions, all
key objectives of relationship marketing.
Our finding that customer affect is driven mainly by the
authenticity of the emotional labor display, rather than the
extent of employee smiling, has important implications for
service marketers. First, it contradicts the common mantra
in service organizations that service with a smile and
friendly service staff inevitably pay off and serve as a
competitive advantage. The commonly advocated “Smile!”
policy that is currently in place in many firms may be a less
useful strategy than is generally believed. Rather, our
results indicate that the authenticity of the emotional dis-
play by frontline staff and the sincerity with which staff
interacts with customers may be much stronger drivers of
service outcomes than policies that require people to smile
at any cost but do not provide them with the tools to create
and display seemingly real emotions when they serve cus-
tomers. Simply hiring low-paid service workers with lim-
ited qualifications and motivation and requiring them to
smile at customers as part of their job description may not
deliver the desired results, whereas the display of authentic
feelings by service employees—facilitated by the use of
deep acting—is likely to be more effective for positively
influencing customer satisfaction and related service
outcomes.
Second, in terms of managing frontline employees, the
provision of authentic feelings is a far more challenging
task than simply smiling at customers. Such strategy may
require increased managerial emphasis on identifying and
hiring talented and qualified frontline employees. Managers
also may need to provide additional training for frontline
staff that teaches them how to engage in deep acting. One
such approach uses a perspective-taking technique, which
puts employees in the customers’ shoes and thereby
increases their ability to adopt a customer’s viewpoint
(Parker and Axtell 2001). For example, flight attendant
trainees can be taught strategies to deal with irate cus-
tomers, such as imagining that something traumatic has
happened in these customers’ lives or by pretending that
they are children (Hochschild 1983).
Third, our results confirm that the emotions customers
experience during service encounters play crucial roles and
directly affect the success of service relationships. Because
customer emotions appear to be key drivers of rapport with
employees and, ultimately, customer satisfaction and loy-
alty intentions, service organizations may benefit from
focusing their attention on increasing positive customer
emotions. This recommendation is consistent with emerg-
ing literature on customer delight (Rust and Oliver 2000),
which stresses the emotional component of customers’ ser-
vice evaluations.

How Emotional Contagion and Emotional Labor Affect Service Relationships / 71
REFERENCES
Anderson, James C. and David W. Gerbing (1988), “Structural
Equation Modeling in Practice: A Review and Recommended
Two-Step Approach,” Psychological Bulletin, 103 (3), 347–62.
Barsade, Sigal G. (2002), “The Ripple Effect: Emotional Conta-
gion and Its Influence on Group Behavior,” Administrative Sci-
ence Quarterly, 47 (December), 644–75.
Bartel, Caroline A. and Richard Saavedra (2000), “The Collective
Construction of Work Group Moods,” Administrative Science
Quarterly, 45 (June), 197–231.
Bitner, Mary Jo (1990), “Evaluating Service Encounters: The
Effects of Physical Surroundings and Employee Responses,”
Journal of Marketing, 54 (April), 69–82.
——— and Amy R. Hubbert (1994), “Encounter Satisfaction Ver-
sus Overall Satisfaction Versus Quality: The Customer’s
Voice,” in Service Quality: New Directions in Theory and
Practice, Roland T. Rust and Richard L. Oliver, eds. Thousand
Oaks, CA: Sage Publications, 72–94.
Brief, Arthur P., Michael J. Burke, Jennifer M. George, Brian S.
Robinson, and Jane Webster (1988), “Should Negative Affec-
tivity Remain an Unmeasured Variable in the Study of Job
Stress?” Journal of Applied Psychology, 73 (May), 193–98.
Brotheridge, Céleste M. and Alicia A. Grandey (2002), “Emo-
tional Labor and Burnout: Comparing Two Perspectives of
‘People Work,’” Journal of Vocational Behavior, 60 (Febru-
ary), 17–39.
Burke, Michael J., Arthur P. Brief, Jennifer M. George, Loriann
Roberson, and Jane Webster (1989), “Measuring Affect at
Work: Confirmatory Analyses of Competing Mood Structures
with Conceptual Linkage to Cortical Regulatory Systems,”
Journal of Personality and Social Psychology, 57 (6),
1091–1102.
Chin, Wynne W. (1998), “The Partial Least Square Approach to
Structural Equations Modeling,” in Modern Methods for Busi-
ness Research, George A. Marcoulides, ed. Mahwah, NJ:
Lawrence Erlbaum Associates, 295–336.
——— (2001), PLS-Graph User’s Guide: Version 3.0. Houston:
Soft Modeling Inc.
APPENDIX B
List of Items and Goodness-of-Fit Measures of
PLS Analysis
Change in Customer Positive Affect (Composite
Reliability [CR] = .871, Average Variance Extracted
[AVE] = .627)
At this moment, I feel …
elated. .794
peppy. .760
enthusiastic. .800
excited. .813
Customer Satisfaction with Transaction (CR = .948,
AVE = .784)
I am delighted by this service experience. .922
This service experience really helped me to find a
good movie. .764
This service experience was a great one. .863
I am satisfied with this specific service experience. .939
I really liked this service experience. .928
Customer–Employee Rapport (CR = .977, AVE = .913)
In thinking about my relationship with this person,
I enjoyed interacting with this employee. .949
This employee created a feeling of “warmth” in our
relationship. .958
This employee related well to me. .947
I was comfortable interacting with this employee. .969
Future Loyalty Intentions (CR = .956, AVE = .846)
The next time I need the services of a video store,
I will choose this company. .936
I will say positive things about this service provider
to other people. .922
I plan to visit this service provider in the next years. .922
I will recommend this service provider to someone
who seeks my advice. .898
Notes: Numbers in the right-hand column are coefficients of
determination.
Step 1. As soon as the customer enters the video consult-
ing store and steps toward the service counter, the
employee makes eye contact with the customer
and welcomes him or her to the store.
Step 2. The employee briefly introduces the nature of the
movie consulting business and its service offerings
following a standardized script. The employee
points out that this service is currently being tested
and thus is free of charge to the customer.
Step 3. The employee asks a series of standardized ques-
tions about the customer’s movie-viewing habits
and preferences. These questions follow a stan-
dardized order and cover the following topics: (1)
preferences for specific movie genres, (2) prefer-
ences for happy endings, (3) importance of film
directors in influencing choice of movies, (4) prefer-
ences for specific film directors, (4) preferences for
specific lead actors and actresses, and (5) prefer-
ences for a movie’s country of origin. During this
process, the employee takes brief notes while the
customer answers all questions.
Step 4. Following the set of standardized questions, the
employee briefly reviews his or her notes and sum-
marizes the customer’s responses.
Step 5. The employee proceeds to recommend a specific
movie to the customer on the basis of the cus-
tomer’s responses. The employee briefly introduces
the movie using a standardized script. If the cus-
tomer has already seen this movie or is not inter-
ested for other reasons, the employee recom-
mends up to two additional movies.
Step 6. When the movie choice is finalized, the employee
asks the customer for his or her choice of DVD or
VHS format, and the employee hands the movie to
the customer. The customer is thanked for his or
her business and leaves the video consulting store.
APPENDIX A
Summary of Service Script

72 / Journal of Marketing, July 2006
Churchill, Gilbert A. and Carol Surprenant (1982), “An Investiga-
tion into the Determinants of Customer Satisfaction,” Journal
of Marketing Research, 19 (November), 491–504.
DeWitt, Tom and Michael K. Brady (2003), “Rethinking Service
Recovery Strategies,” Journal of Service Research, 6 (Novem-
ber), 193–207.
——— and Yongmei Liu (2002), “The Customer Orientation-
Loyalty Model: The Role of Emotional Contagion and Rapport
in the Service Encounter,” in AMA Summer Educators’ Confer-
ence: Enhancing Knowledge Development in Marketing,
William J. Kehoe and John H. Lindgren Jr., eds. Chicago:
American Marketing Association, 321.
Ekman, Paul (1992), “Facial Expressions of Emotions: New Find-
ings, New Questions,” Psychological Science, 3 (January),
34–38.
———, Richard J. Davidson, and Wallace V. Friesen (1990), “The
Duchenne Smile: Emotional Expression and Brain Psychology
II,” Journal of Personality and Social Psychology, 58 (2),
342–58.
——— and Wallace V. Friesen (1982), “Felt, False, and Miserable
Smiles,” Journal of Nonverbal Behavior, 6 (4), 238–52.
———, ———, and Maureen O’Sullivan (1988), “Smiles When
Lying,” Journal of Personality and Social Psychology, 54 (3),
414–20.
Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, Intention,
and Behavior: An Introduction to Theory and Research. Read-
ing, MA: Addison-Wesley.
Fornell, Claes (1992), “A National Customer Satisfaction Barome-
ter: The Swedish Experience,” Journal of Marketing, 56 (Janu-
ary), 6–21.
——— and Fred L. Bookstein (1982), “Two Structural Equation
Models: LISREL and PLS Applied to Consumer Exit-Voice
Theory,” Journal of Marketing Research, 19 (November),
440–52.
——— and David F. Larcker (1981), “Evaluating Structural Equa-
tion Models with Unobservable Variables and Measurement
Error,” Journal of Marketing Research, 18 (February), 39–50.
Geisser, Seymour (1974), “A Predictive Approach to the Random
Effects Model,” Biometrika, 61 (1), 101–107.
Gillis, John S., Frank J. Bernieri, and Eric Wooten (1995), “The
Effects of Stimulus Medium and Feedback on the Judgment of
Rapport,” Organizational Behavior and Human Decision Pro-
cesses, 63 (July), 33–46.
Grandey, Alicia A. (2003), “When ‘The Show Must Go On’: Sur-
face and Deep Acting as Determinants of Emotional Exhaus-
tion and Peer-Rated Service Delivery,” Academy of Manage-
ment Journal, 46 (February), 86–96.
———, Glenda M. Fisk, Anna S. Mattila, Karen J. Jansen, and
Lori A. Sideman (2005), “Is ‘Service With a Smile’ Enough?
Authenticity of Positive Displays During Service Encounters,”
Organizational Behavior and Human Decision Processes, 96
(January), 38–55.
Gremler, Dwayne D. and Kevin P. Gwinner (2000), “Customer-
Employee Rapport in Service Relationships,” Journal of Ser-
vice Research, 3 (August), 82–104.
Grove, Stephen J. and Raymond P. Fisk (1992), “The Service
Experience as Theater,” in Advances in Consumer Research,
Vol. 19, John F. Sherry and Brian Sternthal, eds. Provo, UT:
Association of Consumer Research, 455–61.
Gump, Brooks B. and James A. Kulik (1997), “Stress, Affiliation,
and Emotional Contagion,” Journal of Personality and Social
Psychology, 72 (2), 305–319.
Gwinner, Kevin P., Dwayne D. Gremler, and Mary Jo Bitner
(1998), “Relational Benefits in Services Industries: The Cus-
tomer’s Perspective,” Journal of the Academy of Marketing Sci-
ence, 26 (Spring), 101–114.
Hatfield, Elaine, John T. Cacioppo, and Richard L. Rapson (1994),
Emotional Contagion. Cambridge, UK: Cambridge University
Press.
Hennig-Thurau, Thorsten and Alexander Klee (1997), “The
Impact of Customer Satisfaction and Relationship Quality on
Customer Retention: A Critical Reassessment and Model
Development,” Psychology & Marketing, 14 (8), 737–64.
Hirschman, Albert O. (1970), Exit, Voice, and Loyalty: Responses
to Decline in Firms. Cambridge, MA: Harvard University
Press.
Hochschild, Arlie R. (1983), The Managed Heart: Commercializa-
tion of Human Feeling. Berkeley: University of California
Press.
Hunt, H. Keith (1993), “CSD&CB Research Suggestions and
Observations for the 1990’s,” Journal of Satisfaction/Dissatis-
faction and Complaining Behavior, 6, 40–42.
Kruml, Susan M. and Deanna Geddes (2000), “Exploring the
Dimensions of Emotional Labor: The Heart of Hochschild’s
Work,” Management Communication Quarterly, 14 (1), 8–49.
McHugo, Gregory J., Denis G. Sullivan, John T. Lanzetta, Roger
D. Masters, and Basil G. Englis (1985), “Emotional Reactions
to the Expressive Displays of a Political Leader,” Journal of
Personality and Social Psychology, 49 (6), 1513–29.
Morris, J. Andrew and Daniel C. Feldman (1996), “The Dimen-
sions, Antecedents, and Consequences of Emotional Labor,”
Academy of Management Review, 21 (October), 986–1010.
Oliver, Richard L. (1981), “Measurement and Evaluation of Satis-
faction Processes in Retail Settings,” Journal of Retailing, 57
(Fall), 25–48.
——— (1997), Satisfaction: A Behavioral Perspective on the
Consumer. New York: McGraw-Hill.
——— (1999), “Whence Consumer Loyalty?” Journal of Market-
ing, 63 (Special Issue), 33–44.
Parasuraman, A., Valarie A. Zeithaml, and Leonard L. Berry
(1985), “A Conceptual Model of Service Quality and Its Impli-
cations for Future Research,” Journal of Marketing, 49 (Fall),
41–50.
Parker, Sharon K. and Carolyn M. Axtell (2001), “Seeing Another
Viewpoint: Antecedents and Outcomes of Employee Perspec-
tive Taking,” Academy of Management Journal, 44 (Decem-
ber), 1085–1100.
Perdue, Barbara C. and John O. Summers (1986), “Checking the
Success of Manipulations in Marketing Experiments,” Journal
of Marketing Research, 23 (November), 317–26.
Peter, J. Paul, Gilbert A. Churchill Jr., and Tom J. Brown (1993),
“Caution in the Use of Difference Scores in Consumer
Research,” Journal of Consumer Research, 19 (March),
655–62.
Pugh, S. Douglas (2001), “Service with a Smile: Emotional Conta-
gion in the Service Encounter,” Academy of Management Jour-
nal, 44 (October), 1018–1027.
Rafaeli, Anat and Robert I. Sutton (1987), “Expression of Emotion
as Part of the Work Role,” Academy of Management Review, 12
(January), 23–37.
——— and ——— (1989), “The Expression of Emotion in Orga-
nizational Life,” in Research in Organizational Behavior, Vol.
11, Larry L. Cummings and Barry M. Staw, eds. Greenwich,
CT: JAI Press, 1–42.
Rust, Roland T. and Richard L. Oliver (2000), “Should We Delight
the Customer?” Journal of the Academy of Marketing Science,
28 (Winter), 86–94.
——— and Anthony J. Zahorik (1993), “Customer Satisfaction,
Customer Retention, and Market Share,” Journal of Retailing,
69 (Summer), 193–215.
Salancik, Gerald R. and Jeffrey Pfeffer (1978), “A Social Informa-
tion Processing Approach to Job Attitudes and Task Design,”
Administrative Science Quarterly, 23 (June), 224–53.

How Emotional Contagion and Emotional Labor Affect Service Relationships / 73
Schoenewolf, Gerald (1990), “Emotional Contagion: Behavior
Induction in Individuals and Groups,” Modern Psychoanalysis,
15 (1) 49–61.
Schwartz, Norbert and Gerald L. Clore (1988), “How Do I Feel
About It? The Informative Function of Affective States,” in
Affect, Cognition, and Social Behavior, Klaus Fiedler and
Joseph P. Forgas, eds. Toronto: Hogrefe, 44–62.
Shrout, Patrick E. and Joseph L. Fleiss (1979), “Intraclass Correla-
tions: Uses in Assessing Rater Reliability,” Psychological Bul-
letin, 86 (2), 420–28.
Stanislavski, Constantin (1965), An Actor Prepares. New York:
Theatre Art Books.
Stone, Mervyn (1974), “Cross-Validatory Choice and Assessment
of Statistical Predictions,” Journal of the Royal Statistical Soci-
ety, 36 (2), 111–47.
Surprenant, Carol F. and Michael R. Solomon (1987), “Pre-
dictability and Personalization in the Service Encounter,” Jour-
nal of Marketing, 51 (April), 86–96.
Sutton, Robert I. and Anat Rafaeli (1988), “Untangling the Rela-
tionship Between Displayed Emotions and Organizational
Sales: The Case of Convenience Stores,” Academy of Manage-
ment Journal, 31 (September), 461–87.
Taylor, Steven A. and Thomas L. Baker (1994), “An Assessment of
the Relationship Between Service Quality and Customer Satis-
faction in the Formation of Consumers’ Purchase Intentions,”
Journal of Retailing, 70 (Summer), 163–78.
Tickle-Degnen, Linda and Robert Rosenthal (1990), “The Nature
of Rapport and Its Nonverbal Correlates,” Psychological
Inquiry, 1 (4), 285–93.
Tisak, John and Carla S. Smith (1994), “Defending and Extending
Difference Score Methods,” Journal of Management, 20 (3),
675–82.
Tsai, Wei-Chi (2001), “Determinants and Consequences of
Employee Displayed Positive Emotions,” Journal of Manage-
ment, 27 (4), 497–512.
——— and Yin-Mei Huang (2002), “Mechanisms Linking
Employee Affective Delivery and Customer Behavioral Inten-
tions,” Journal of Applied Psychology, 87 (October),
1001–1008.
Verbeke, Willem (1997), “Individual Differences in Emotional
Contagion of Salespersons: Its Effect on Performance and
Burnout,” Psychology & Marketing, 14 (6), 617–36.
Watson, David and Auke Tellegen (1985), “Toward a Consensual
Structure of Mood,” Psychological Bulletin, 98 (2), 219–35.
Yi, Youjae (1990), “A Critical Review of Consumer Satisfaction,”
in Review of Marketing 1990, Valarie A. Zeithaml, ed. Chicago:
American Marketing Association, 68–123.
Zeithaml, Valarie A., Leonard L. Berry, and A. Parasuraman
(1996), “The Behavioral Consequences of Service Quality,”
Journal of Marketing, 60 (April), 31–46.