Content uploaded by Jeremy D. Mackey
Author content
All content in this area was uploaded by Jeremy D. Mackey on Jul 23, 2019
Content may be subject to copyright.
ORIGINAL EMPIRICAL RESEARCH
Employee satisfaction trajectories and their effect on customer
satisfaction and repatronage intentions
Jeremy S. Wolter
1
&Dora Bock
1
&Jeremy Mackey
2
&Pei Xu
3
&Jeffery S. Smith
4
Received: 13 December 2017 /Accepted: 17 April 2019
#Academy of Marketing Science 2019
Abstract
Does improving employee happiness affect customer outcomes? The current study attempts to answer this question by examining the
impact of employee satisfaction trajectories (i.e., systematic changes in employee satisfaction) on customer outcomes. After accounting
for employees’initial satisfaction levels, the analyses demonstrate the importance of employee satisfaction trajectories for customer
satisfaction and repatronage intentions, as well as identify customer-employee contact as a necessary conduit for their effect. From a
macro perspective, employee satisfaction trajectories strongly impact customer satisfaction for companies with significant employee–
customer interaction, but not for companies without such interaction. From a micro perspective, employee satisfaction trajectories
influence customer repatronage intentions for frequent customers, but not for infrequent customers. These effects are robust to
controlling for previous customer evaluations and recent employee evaluations. Overall, these findings extend the dominant view of
examining static, employee satisfaction levels and offer important implications for the management of the organizational frontline.
Keywords Employee satisfaction trajectory .Customer satisfaction .Latent growth curve .Organizational frontline
Over the past two decades, research has firmly established an
association between employee satisfaction and customer sat-
isfaction (Bitner 1995; Brown and Lam 2008;Harteretal.
2002; Hogreve et al. 2017). Implicit in this association is that
firms’investments to increase employee satisfaction spill over
to customers and favorably influence customer satisfaction
(e.g., Hogreve et al. 2017; Weber 2016). Yet, it is unclear if
this fundamental assumption is true, as research has primarily
adopted a cross-sectional approach to studying employee–
customer associations. In a cross-sectional approach, re-
searchers compare employee satisfaction Blevels^with cus-
tomer satisfaction Blevels^across organizations, subunits, or
employees (e.g., Evanschitzky et al. 2012; Wangenheim et al.
2007). Problematically, a Blevels^(i.e., static) perspective,
which captures employee satisfaction at one point in time,
does not reveal a complete account of the employee–
Michael Ahearne served as Area Editor for this article.
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s11747-019-00655-9) contains supplementary
material, which is available to authorized users.
*Jeremy S. Wolter
jswolter@auburn.edu
Dora Bock
deb0022@auburn.edu
Jeremy Mackey
jdm0096@auburn.edu
Pei Xu
pzx0002@auburn.edu
Jeffery S. Smith
jsmith74@vcu.edu
1
Raymond J. Harbert College of Business, Department of Marketing,
Auburn University, 405 W. Magnolia Ave, Auburn, AL 36849, USA
2
Raymond J. Harbert College of Business, Department of
Management, Auburn University, Auburn, AL, USA
3
Raymond J. Harbert College of Business, Department of Systems
and Technology, Auburn University, Auburn, AL, USA
4
School of Business, Department of Supply Chain and Analytics,
Virginia Commonwealth University, P.O. Box 844000,
Richmond, VA 23284-4000, USA
Journal of the Academy of Marketing Science
https://doi.org/10.1007/s11747-019-00655-9
customer satisfaction relationship. A levels perspective con-
flates between-unit and within-unit variance such that one
does not know if perceived differences from employee satis-
faction are simply a result of fundamental differences between
the units (Certo et al. 2016). Stated differently, comparing an
organization with high employee satisfaction to one that is low
does not guarantee that enhancing the low organization’s
scores will favorably influence customer satisfaction.
Even if changes in employee satisfaction are consequential for
customer satisfaction, it is unknown how such changes compare
to the relative level of employee satisfaction. Consider the fol-
lowing example (adapted from Liu et al. 2012). Employee A’s
satisfaction rises from 2 to 3 and employee B’s satisfaction falls
from 5 to 4. Which employee will favorably affect customer
satisfaction the most? The levels approach suggests that employ-
ee B will have the strongest positive impact on customer satis-
faction because his or her initial satisfaction (5), average satisfac-
tion (4.5), and final satisfaction (4) are all higher than employee
A’s scores. However, incorporating the direction of the change
offers a different perspective. Because of declining satisfaction,
employee B may begin to express frustrations, withdraw from
work, and reduce customer-oriented behaviors. On the other
hand, employee A’s trajectory may result in the opposite effect,
such that with an improving satisfaction, employee A may per-
ceive his or her job is improving, may make greater effort in daily
work performance, and may display even more affective cues to
others (Chen et al. 2011;Liuetal.2012).
Given the unverified assumption that changes in employee
satisfaction affect customer satisfaction and the possibility that
such change is more relevant than the level, research needs to
develop theory that pertains to how changes in employee satis-
faction operate and affect customer satisfaction. The current
research’s primary purpose, then, is to take an initial step in
developing and testing such theory. As illustrated in Fig. 1,we
introduce the concept of employee satisfaction trajectory,defined
as systematic changes or trends in employees’satisfaction over
time,tothemarketingdomain(Liuetal.2012). Drawing from
the emotional contagion literature and the emotions as social
information model (Van Kleef 2009), we theorize and test how
employee satisfaction trajectories influence customer satisfaction
over employee satisfaction levels and the extent to which such
effects are dependent on employee–customer contact.
The current research examines employee satisfaction trajecto-
ries across two studies utilizing latent growth modeling on lon-
gitudinal data of employees. The first study provides a macro
perspective that examines employee satisfaction levels and tra-
jectories across firms, as well as their subsequent impact on
customer satisfaction. The results of this study reveal that the
degree of contact between employees and customers moderates
the effects of satisfaction level and trajectory on customer satis-
faction. Whereas the effect of employee satisfaction level on
customer satisfaction is facilitated by any amount of contact,
the effect of the trajectory is only affected by high contact.
The second study takes a micro perspective by examining
employee satisfaction trajectories across multiple operational
units within a single firm to test a contact hypothesis based on
customer patronage behavior. This study also expands the
dependent variable beyond customer satisfaction to a more
holistic examination of customer repatronage intentions and
operational unit financial metrics. The results reveal that
changes in employee satisfaction strongly affect customers’
repatronage intentions but only for those customers who pa-
tronize frequently. However, the effect of employee satisfac-
tion level is significant for all customers, regardless of their
patronage frequency. Subsequently, repatronage intentions
mediate the effect of employee satisfaction trajectory on future
revenue.
These findings offer several important contributions to the
literature. First, we demonstrate that employee satisfaction
trajectories provide valuable insight from that of employee
satisfaction levels. This finding enhances the existing empiri-
cal literature on changes in employee satisfaction (e.g., Liu
et al. 2012) and expands the nascent but growing field of
dynamics in marketing relationships (e.g., Palmatier et al.
2013) by linking changes in employee attitudes to customers.
Second, our findings identify high employee–customer con-
tact as a greater necessity for employee satisfaction trajectory
than for employee satisfaction level. Thus, the current research
demonstrates an important condition that determines whether
employee satisfaction levels or trajectories are likely to be the
predominant predictor of customer satisfaction and
repatronage. Finally, our finding supports the assertion of
Certo et al. (2016) that valuable contributions may come from
longitudinal research that can assess differences attributable to
within-unit variance from between-unit variance. We expand
upon these contributions in the discussion section.
Conceptual development
The importance of trend phenomena
The plentitude of research linking employee and customer
satisfaction has almost exclusively been conducted from a
Blevels^perspective. The little research that has examined
changes in employee satisfaction does not suggest that such
changes affect customer satisfaction. Table 1lists empirical
research that has examined the effect of changes in employee
satisfaction on customer satisfaction. Of the five articles listed,
four suggest that changes in employee satisfaction either have
no effect or a negative effect on customer satisfaction. Yet,
research on trend phenomenon suggest that changes in em-
ployee satisfaction should be consequential.
Studies show that individuals frequently engage in
mental time travel where they recollect the past and an-
ticipate the future (Killingsworth and Gilbert 2010;Ye
J. of the Acad. Mark. Sci.
et al. 2018). Indeed, the degree to which individuals en-
gage in intertemporal judgments is so pervasive that Hsee
and Abelsen (1991, p. 341) contend that as Breflected in a
number of social psychological theories^,people’ssatis-
faction with any outcome, be it their salary, weight, or
friendship, is a positive function of the change relation-
ship. Research in marketing testifies that customers use
trends as heuristics for understanding their environment.
For example, Palmatier et al. (2013) and Huang and
Cheng (2016) suggest trends in customer commitment
and customer-company identification (i.e., velocities) in-
fluence important firm outcomes, such as sales perfor-
mance and customer loyalty, beyond their associated static
or level measures. Research further suggests that con-
sumers utilize trend-based fallacies and not only find
strengthening performances attractive, but also wrongly
expect such trends to continue (Johnson et al. 2005).
Similarly, service customers are cognizant of employees’
service recovery progress or the lack thereof (Marinova
et al. 2018). In fact, customer defection is often a reluc-
tant choice made after a culmination of negative events
that suggests a deteriorating trend in the relationship
(Hollmann et al. 2015).
Employees also use trends to understand their environ-
ment. For example, Ye et al. (2007) utilize schema theory
as well as assimilation-contrast theory to suggest that
frontline employees can become psychologically detached
when they perceive their organization is continually
changing from the pursuit of cost containment. Li et al.
(2017) apply gestalt characteristics theory to suggest that
employees’perceptions of trends in their job complexity
affect job strain. Furthermore, declining commitment to
an organization is worse for generating turnover than
consistently low commitment, which suggests employees
are perceptive to these trends (Bentein et al. 2005).
Of particular importance to the current research is the
realization that employees experience trends in their sat-
isfaction and that such trends can be a stronger predictor
of employee turnover than the static satisfaction level
(Chen et al. 2011;Liuetal.2012). In accord with spirals
theory,thetrendinanemployee’s satisfaction can affect
that employee’s work expectations for the future (Chen
et al. 2011). As a result, increasing positive or negative
work experiences promotes the belief that these experi-
ences will continue to get either better or worse.
Employee satisfaction follows this belief in a self-
fulfilling prophecy (Hobfoll 1989; Kahneman and
Tversky 1979).
In summary, research across disciplines suggest both cus-
tomers and employees are observant and reactive to trends in
relation to their organizational relationships. Though existing
research generally indicates that changes in employee satis-
faction trajectories fail to cross the organizational frontline to
affect customer satisfaction (see Table 1), the wide variety of
recent research on marketing related trend phenomenon sug-
gests an effect should exist. This discrepancy suggests the
presence of an unidentified moderator, overlooked in previous
research, raising the question—in what conditions may chang-
es in employee satisfaction affect customer satisfaction?
Research suggests employee–customer contact is one of the
strongest moderators of the employee satisfaction and custom-
er satisfaction link (Brown and Lam 2008), which is notable
because the research listed in Table 1does not account for
such contact. Thus, we anticipate that employee–customer
contact serves as a determining factor facilitating the transfer-
ence of satisfaction from employees to customers. To address
Customer satisfaction
Study 1
Value perceptions
Repatronage intentions
Study 2
The modera ting role of
employee -cu stome r
contact
Employe e
mindset metrics
Cust omer mind set
metrics
Futur e re venue
Study 2
Financial
metrics
As the number of service encounters
increases, the encounter takes on more
importance, increasing the amount of
attention and elaboration of cues and
pushi ng c ustome rs towa rds contr astin g
subsequent encounters against former
ones.
Relevant theory: Inferential processes
within the EASI (e.g., disconfirm ation)
The service encounter provides
customers cues to employees’ internal
states and facilitates subconscious
emotional contagion and mimicry
(Hoegreve et al 2016).
Relevant theory: Affective processes
within the EASI (e.g., e motional
contagion)
Employee
satisfaction level
Employee
satisfaction trajectory
Fig. 1 Illustration of the conceptual model. Notes: Grey boxes represent theoretical processes. White boxes represent measured constructs
J. of the Acad. Mark. Sci.
Table 1 Empirical research on changes in employee satisfaction affecting customer satisfaction
Authors Context Employee construct Customer construct Unit of analysis Empirical approach Theoretical foundation Findings
Ryan et al. (1996) Retail bank
branches
Employee morale
(n ~ 1513)
Customer satisfaction
(n= 1000)
Branch
(n= 131)
Cross-lagged
analysis in SEM
Attraction-selection-
attrition model
Customer satisfaction weakly
predicts change in employee
morale 1 year later. The reverse
effect, in which employee morale
predicts changes in customer
satisfaction, is not supported.
Scharitzer and
Korunka (2000)
Two c ust om er
service centers
Employee satisfaction
(n= 104)
Customer satisfaction
(n=902)
Employee
&customers
separately
Means analysis Total quality management Average employee satisfaction
decreased over time while average
customer satisfaction increased
over time.
Keiningham
et al. (2006)
Large specialty
retailer
Employee satisfaction
(n> 3900)
Customer satisfaction
(n> 34,000)
Retail centers
(n= 120)
Correlation analysis
using difference
scores
Service profit chain,
satisfaction mirror,
Kano’stheoryof
attractive quality
Changes in employee satisfaction
exhibited a negative relationship
with changes in customer satisfaction.
Evanschitzky
et al. (2012)
Large DIY
retail chain
Employee satisfaction
(n= 16,463)
Customer satisfaction
(n= 407,238)
Retail centers
(n= 119)
3SLS regression Service profit chain Employee satisfaction predicts
synchronous customer satisfaction
over previous year’s customer satisfaction.
Zablah et al. (2016) Specialty apparel
retailer
Employee satisfaction
(n= 1470)
Customer satisfaction
(n= 49,272)
Employee–
customer dyad
(n= 1470)
Cross-lagged
analysis in SEM
Satisfaction mirror Customer satisfaction predicts change
in employee satisfaction 1 year later
but not vice-versa. Customer
engagement increases the
outside-in effect.
J. of the Acad. Mark. Sci.
this possibility, we first review how employee satisfaction and
customer satisfaction are intrinsically linked through emotion-
al processes and then how contact facilitates these processes.
From employee satisfaction to customer satisfaction:
The power of emotional displays
Though there are many theoretical frameworks that link
employee actions to customer evaluations (e.g., service
profit chain), emotional contagion is a primary theory
for studying the relationship between employee satisfac-
tion and customer satisfaction within limited (as op-
posed to relational) service encounters (Brown and
Lam 2008;HomburgandStock2004). Emotional con-
tagion builds off the notion that emotions are expressed
in social interaction and suggests a more pleasurable
customer experience results from affect transfer, where-
by a satisfied employee expresses positive emotions,
either verbally or non-verbally, that are contagious to
customers (Peters and Kashima 2015; Pugh 2001).
That is, employee actions are reflective of their emo-
tional state, and customers Bcatch^these displayed emo-
tions. Several theoretical explanations, including subcon-
scious and conscious processes, account for the occur-
rence of emotional contagion (c.f. Peters and Kashima
2015).
However, emotional transfer is unlikely to account for the
effect of employee satisfaction trajectories on customers as
perceiving change (e.g., mental time travel) is a cognitive
process. Thus, we utilize the emotions as social information
(EASI) model because it is a unifying framework that ac-
counts for when and how emotional expressions by one party
affect the behavior in another party through both affective and
cognitive means (Van Kleef 2009). Specifically, the EASI
model posits that an expresser’s emotional display affects an
observer’s behavior through two possible processes: affective
or inferential. For an affective reactions process, the expressed
emotions affect the observer’s emotions (i.e., emotional con-
tagion) and/or liking of the expresser. For an inferential pro-
cess, the observed emotional display provides relevant infor-
mation about the situation to the observer, leading the observ-
er to make inferences about oneself, the expresser, or other
targets associated with the situation (Van Kleef 2009). In the
subsequent sections, we draw upon this model and the
existing marketing literature to explain why contact serves
as the key conduit for the effects of employee satisfaction
levels and trajectories.
Employee satisfaction levels and employee–customer
contact
Employee–customer contact plays a critical role in emotional
contagion processes and the linkage between employee and
customer satisfaction. Applying the EASI model to the cus-
tomer context suggests that within instances of employee–
customer contact, employee displays of satisfaction can lead
to customer satisfaction through customer emotions, liking
(i.e., affective reactions pathway) or inferences (i.e., inferen-
tial pathway). This implies that an employee’s positive emo-
tional display might make the customer happy and like the
employee (i.e., experience an affective reaction), or it might
initiate a series of inferences (e.g., employee satisfaction, ex-
cellent service, good company). Ultimately, the customer’s
affective reactions and/or inferences motivate the customer
to be satisfied and continue frequenting the firm.
Consistent with the interpersonal nature of emotions posit-
ed by EASI model, studies in marketing show that contact,
even limited contact, can facilitate affective reactions and in-
ferential pathways. Table 2shows quantitative research that
studies the effect of employee–customer contact on the rela-
tionship between employee and customer affective states.
Here, the research suggests that a business unit’sorcompany’s
employees need to have contact with customers for employee
satisfaction (or service climate) to correspond with customer
metrics (Brown and Lam 2008; Dietz et al. 2004; Mayer et al.
2009). Considering differences in contact across customers
(such as by varying amounts of patronage), even infrequent
contact on a customer’s part facilitates the correspondence
between employee and customer satisfaction, at least from a
levels perspective (Homburg and Stock 2004). In other words,
if customers have some contact with employees in a service
that necessitates face-to-face interactions and the observation
of service production, employees’happiness can be conta-
gious and facilitate better service that customers notice.
Thus, consistent with extant research, we anticipate that
employee–customer contact strengthens the relationship be-
tween employee satisfaction and customer satisfaction.
H1: The effect of employee satisfaction level on customer
satisfaction is dependent on employee–customer contact
such that the effect of the level strengthens as employee–
customer contact increases.
Employee satisfaction trajectories
and employee–customer contact
If contact is necessary for employee satisfaction levels to affect
customer satisfaction, then certainly contact is required for em-
ployee satisfaction trajectories as well. However, unlike levels,
trajectories require observation and comparison between states
to be fully impactful. That is, the relationship between employ-
ee satisfaction trajectories and customer satisfaction is more
nuanced than the relationship between employee satisfaction
levels and customer satisfaction, as trajectories primarily oper-
ate through inferential processing. With trajectories, em-
ployee’s emotional displays and behavior not only need to be
J. of the Acad. Mark. Sci.
Table 2 Empirical research on the effect of employee–customer contact on the relationship between employee and customer metrics
Authors Context Employee construct Customer
construct
Unit of analysis Empirical
approach
Theoretical
foundation
Findings
Dietz et al. (2004) Retail bank Service climate
(n= 2616)
Customer
satisfaction
(n = 17,480)
Branches (retail banks)
(n= 160)
Regression with
interaction term
Service climate Branches with higher average frequency
of customer patronage have a stronger
relationship between service climate
and customer satisfaction. At low
levels of average frequency, the effect
of service climate is not significant.
The range of contact includes no
visits in the last 6 months.
Homburg and Stock
(2004)
Manufacturers, banks,
and insurance
providers
Job satisfaction
(n= 1305)
Customer
satisfaction
(n=222)
Employee–customer
dyad
(n= 111)
Multiple groups
SEM
Emotional contagion The more a customer frequents a service,
the stronger the effect of job
satisfaction on customer satisfaction.
At low levels of contact, the effect of
employee satisfaction is small but
significant. Contact ranges from one
visit in the past year to daily visits.
Wangenheim et al. (2007) DIY retail chain Employee satisfaction
(n= 1659)
Customer
satisfaction
(n= 53,645)
Employee (n= 1659) Multiple groups
SEM
Attraction
selection model
The effect of employee satisfaction on
customer satisfaction is not statistically
different across employee groups
with different levels of customer
contact. The range of contact
includes employees with no contact.
Brown and Lam (2008) N/A Employee satisfaction Customer
satisfaction
N/A Meta-analytic Service profit chain The effect of employee satisfaction
on customer satisfaction is
dependent on whether a service
requires up close interactions that
allow customers to observe service
production but not on whether a
service requires ongoing contact
because it is relational. Thus, the
moderating effect of contact is most
pronounced when transitioning
through lower levels.
Groth et al. (2009)Unspecifiedservice
industries
Deep and surface
acting
(n=285)
Service quality
(n=285)
Employee–customer
dyad
(n= 285)
PLS with
interaction term
Emotional labor The deep acting ability of employees has the
same effect on perceived service quality
regardless of differences in contact with
customers. The range of contact spans
from moderate to high.
Mayer et al. (2009) Supermarket chain Service climate
(n=4500)
Customer
satisfaction
(n= 14,835)
Department (n=129) Regressionwith
interaction term
Service climate A department’samountofcontactwith
customers moderates the effect of service
climate on customer satisfaction. At low
levels of contact, the effect of service
climate is insignificant. The range of
contact includes departments with no
contact.
J. of the Acad. Mark. Sci.
noticeably different than before, but customers must also be
observant to this change and to infer its meaning. In the EASI
model, this type of influence represents inferential processing
because customers must take notice of the change in the em-
ployees or the performance provided and realize its personal
significance (Van Kleef 2009). Consequently, for the inferential
pathway to be fruitful, customers must have the motivation and
ability to process the change in employees’displays.
Importantly, the nature of contact, especially repeated contact,
fosters situational involvement that pushes customers to be
attentive to cues that become the evidence of an employee
satisfaction trajectory. That is, contact creates customer moti-
vation and provides the capability for customers to observe and
react to changes in employee satisfaction.
Within the marketing literature, empirical evidence aligns
with the expectation that contact facilitates the inferential
route posited by the EASI model. Specifically, research on
contrast effects suggests that repeated contact creates situa-
tional involvement and a motivation to engage in effortful
processing that leads customers to be attentive to discrepan-
cies between states (Babin et al. 1994; Meyers-Levy and
Sternthal 1993; Petty et al. 1983). In instances of employee–
customer contact, the personal relevance of the service expe-
rience (i.e., situational involvement) increases as customers
participate in co-production (Mende and Van Doorn 2014).
In other words, when customers have repeated contact with
employees, the customers become more involved in the ex-
changes, more cognizant of employees dispositional states,
and more discerning of when those states have changed,
which enables customers to make inferences that changes with
personal significance are occurring (i.e., customers engage in
the inferential pathway). Repeated contact, then, not only pro-
vides the means by which emotional contagion and the service
chain operate, as the research in Table 2attests, but it also
provides the environment to observe changes and the motiva-
tion to attend to cues of employees’change. Therefore, we
expect that employee–customer contact also strengthens the
relationship between changes in employee satisfaction and
customer satisfaction.
H2: The effect of employee satisfaction trajectories on custom-
er satisfaction is dependent on employee–customer con-
tact such that the effect of trajectories strengthens as
employee–customer contact increases.
The degrees of contact required by employee
satisfaction levels versus trajectories
Though employee–customer contact should strengthen the
relationship between customer satisfaction and both em-
ployee satisfaction level and trajectory, we expect the
moderating effect of contact to slightly differ for the level
and trajectory of employee satisfaction. Within brief
encounters, employee satisfaction levels can affect cus-
tomer satisfaction through emotional contagion (e.g.,
mimicry) processes (i.e., the affective reactions pathway).
In fact, it has been claimed that Bhuman beings can and
will ‘catch’the emotional state of others with whom they
interact or simply share space^(Segrin 2004,p.836).
Though contact facilitates the transfer of affect, only a
limited amount of contact is needed for customers to ob-
serve and be influenced by displayed emotions (and there-
by employee satisfaction levels).
Furthermore, the EASI model suggests that norms are a
moderating factor of affective reactions to emotional dis-
plays. Individuals have Bstronger affective reactions to
inappropriate emotional displays^, and conversely, weak-
er affective reactions to appropriate or expected emotional
displays (Van Kleef 2009, p. 187). Thus, with greater
contact, customers become familiar with emotional dis-
play norms, as represented by employee satisfaction level,
and less affected by such displays. In other words and as
illustrated in Fig. 2by the dotted line, once initial contact
facilitates emotional contagion processes, the effect of
greater contact on emotional contagion should plateau be-
cause customers become accustomed to employees’work-
place happiness. The research shown in Table 2further
supports this idea as studies that examined ranges that
included no contact showed strong effects (Dietz et al.
2004; Homburg and Stock 2004; Mayer et al. 2009)
whereas studies that examined ranges of high contact
showed little or no effect (Brown and Lam 2008;Groth
et al. 2009; Wangenheim et al. 2007).
In addition, the EASI model specifies that as an ob-
server’smotivationandabilitytoprocessinformationcon-
veyed in emotional expressions increase, the inferential
pathway, rather than the affective reaction pathway,
strengthens in predicting that observer’s reactions.
Trajectories require customer attention, and attention to
others is positively influenced by one’s sensory capacities
(Bandura 1971) and situational involvement (Celsi and
Olson 1988), both of which are augmented with increases
in employee–customer contact. Substantial contact
heightens customers’awareness of cues of employees’
changing affective state (e.g., eye contact), allowing small-
er changes to disconfirm expectations. Thus, through con-
tact, the cues manifesting from the trajectory in an em-
ployee’s satisfaction become meaningful to a customer
and the disconfirming evidence of such cues affect the
customers’evaluation of service performance and subse-
quent satisfaction. In other words and as illustrated in Fig.
2by the dashed line, frequent contact provides a sufficient
amount of opportunities and motivation to observe, dis-
cern, and evaluate the trajectory of employees’satisfaction.
In summary, as job satisfaction improves or declines, an
employee invests strategically into or away from his or her
J. of the Acad. Mark. Sci.
work tasks, which manifests in ways that a situationally-
involved customer can perceive. Such cues are more difficult
to discern and more likely to be overlooked than those related
to an employee’s satisfaction level. As such, only high contact
helps facilitate the influence of an employee’s satisfaction tra-
jectory on customer satisfaction. In contrast, even limited con-
tact facilitates the influence of an employee’s satisfaction level
on a customer’s satisfaction and high contact should margin-
ally increase this effect. Thus,
H3a: The moderation of the employee satisfaction level and
customer satisfaction relationship is strongest at low
levels of employee–customer contact rather than high
levels of contact.
H3b: The moderation of the employee satisfaction trajectory
and customer satisfaction relationship is strongest at
high levels of employee–customer contact rather than
low levels.
Study 1
Data structure and variable operationalization
As shown in Fig. 1, Study 1 tests the proposed model from a
macro perspective by examining the effect of employee satis-
faction trajectories on customer satisfaction across companies
using multi-sourced data. The employee data comes from
Large effect
of employee
sasfacon
on customer
sasfacon
No employee -
customer
contact
No effect of
employee
sasfacon
on customer
sasfacon
High employee -
customer
contact
Emoonal contagion
requires at least some
contact to occur.
Upon inial contact,
customers are very
suscepble to the
moods and affect of
employees.
At high levels of contact, general
norms of emoonal display are
established. The effect of similar
emoonal displays does not increase
but, instead, levels off.
At low levels of contact, customers lack
the ability and movaon to observe
trends in employees’ sasfacon.
As contact increases,
customers become movated
observers of employee
sasfacon trajectories.
Current employee
sasfacon states are
compared and contrasted to
previous states. Discrepancies
become parcularly
noteworthy.
Low employee -
customer
contact
Level: β = 0.94 β = 7.94** β= 10.38**
Trajectory: β= 1.58 β= -3.61 β= 35.07**
NA γ= 0.09* γ= 0.10**
NA γ= 0.66 γ= 1.65*
Study 1: Companies with
no contact
Companies with a
lile contact
Companies with a
lot of contact
Level:
Trajectory:
Study 2: For customers who
patronize twice in 2 years
For customers who patronize
six mes in 2 years
NA
Fig. 2 Employee–customer contact moderating the effects of employee
satisfaction level and trajectory based on the Emotions as Social
Information (EASI) Model. Notes: The effects shown are direct effects
of the employee satisfaction construct (level or trajectory) on the depen-
dent variable. These were estimated using a spotlight analysis by center-
ing the contact variable at the desired level of contact
J. of the Acad. Mark. Sci.
Glassdoor.com, a website where employees leave reviews
about the companies for which they work. The company
provided access to 4 years of data, from 2011 to 2014. The
customer data comes from the American Customer
Satisfaction Index (ACSI), which has been used and
validated in many marketing studies (e.g., Anderson et al.
1997). The ACSI score represents a weighted average of three
survey questions (e.g., What is your overall satisfaction with
[our product or service], anchored by Very dissatisfied / Very
satisfied on a 100 point scale). We chose the year (2015) after
the final year of the employee data (2014) for the dependent
variable, which is standard in research connecting employee
and customer satisfaction (Zablah et al. 2016).
1
Because the
ACSI measures satisfaction at the brand level, some compa-
nies had multiple ACSI scores. These multiple scores within a
single company were aggregated together to create a single
satisfaction score similar to Morgan and Rego (2006). There
were 342 unique brands available in the ACSI data as of 2015
that matched with 293 unique firms. The number of employee
reviews ranged from 12 to 8080 for the companies in the final
analysis (as explained below), with an average of 776.
Glassdoor.com measures employee satisfaction by six
categories in which employees evaluate their organization
using 1 to 5 stars. Five of these categories were available
across the entire time of the data: employees’ratings for
wages, leadership, work/life balance, career opportunities,
and overall satisfaction toward their employer. These catego-
ries mirror the pay, supervision, coworkers, work, promotion
subscales of the Job Descriptive Index, which is an extensive-
ly used and validated measure of employee satisfaction
(Kinicki et al. 2002). We combine all five items together as
a measure of employee satisfaction. The items exhibit high
correlations with each other (.50 ≤r≤.76) and reliability
(α= .90). Though Glassdoor allows ex-employees to fill out
a survey, recent research provides predictive validity of this
measure by linking it to firm financial performance (Huang
et al. 2015; Melián-González et al. 2015). We conducted a
non-response assessment, following the procedure of
Palmatier et al. (2013), and found little practical significance
in the analysis (which is reported in full in Web Appendix A).
The Glassdoor data had to be aggregated to the company to
match the ACSI data. Because the ACSI collects its data
across industries throughout the year, we aggregated the
Glassdoor data to match this sequence across the 4 years under
observation. We examined the degree of variance attributable
to the companies in the employee reviews through intraclass
correlation coefficient (ICC) statistics. Across the years, the
variance of employee satisfaction attributable to the
companies ranged from 12 to 17%. Our minimum observed
variance (12%) is deemed suitable for aggregation; therefore,
we aggregated the employee review data to the company level
(Raudenbush and Bryk 2002). Other reliability and agreement
statistics calculated within each time period further confirmed
the appropriateness of aggregation (ICC1 = .12–.15,
ICC2 = .53–.79, Lebreton and Senter 2008).
Three raters assigned the companies to represent levels
of employee–customer contact. Because companies within
an industry were similar in regard to contact, the rating
was conducted at the industry level based on industry
classifications used in the ACSI. Each rater answered an
item derived from Mayer et al. (2009)thatmeasuresthe
degree of employee–customer contact (BHow often do em-
ployees and customers of this company interact face-to-
face as a normal part of consumption?^). Specifically,
raters indicated whether employees and customers interact
rarely or little (coded as 0 for low contact) or interact
often or continuously (coded as 1 for high contact) as a
normal part of consumption for all industries represented
in the data set. An ICC based on the initial rater data
suggests 82% agreement among the raters, which is con-
sidered strong agreement (Lebreton and Senter 2008).
Discrepancies were reexamined and settled based on dis-
cussion among the researchers. A total of 127 companies
(e.g., physical retail, hotels, airlines) were labeled as high
employee–customer contact and 166 companies were la-
beled as low employee–customer contact (e.g., online re-
tailers, energy companies, cable providers, manufacturers).
The full company assignment is available in Web
Appendix B and the correlation matrix is available in
Appendix Table 5.
Analytical technique
To estimate employee trajectories, we utilized Mplus v7.11 to
conduct latent growth modeling in the structural equation
modeling (SEM) framework. Though there are other methods
for longitudinal data, including multilevel models using re-
gression and panel analysis (Zheng et al. 2014), we utilized
latent growth in SEM because of the following reasons. First,
latent growth analysis quantifies a trend over time in employ-
ee satisfaction whereas fixed effects panel analysis ignores
Bthe inherent growth of a variable...because this approach
codes time as a label and overlooks the order or continuity
of time^(Zheng et al. 2014, p. 549). Second, latent growth
modeling in SEM, as compared to multilevel modeling or
random effects panel analysis, estimates indirect effects
(Echambadi et al. 2006), which we utilized in the second
study.
We first estimated the within and between firm variance in
employee satisfaction over time using an ICC statistic based
on an intercept-only model (Raudenbush and Bryk 2002).
1
The use of 2014 ACSI scores rather than the 2015 scores results in no
differences in hypothesis testing though the coefficients for employee satisfac-
tion trajectory and levelon customer satisfaction are smaller for the 2014 ACSI
scores as compared to the 2015 scores.
J. of the Acad. Mark. Sci.
This statistic suggested that 40% of the variance in employee
satisfaction over time was between firms, whereas the remain-
ing variance was within firms. The large within firm variance
is illustrated in Fig. 3, which shows the employee satisfaction
of ten randomly selected companies over the 4 years. Though
some companies’satisfaction is relatively stable and differen-
tiated, such as Little Caesar’s, many companies’employee
satisfaction fluctuates greatly. The significant presence of
within firm variance suggests the trajectory of employee sat-
isfaction, which summarizes the within firm variance, is need-
ed to fully understand the effect of employee satisfaction on
customer satisfaction (Certo et al. 2016).
Some companies (n= 31, 10%) did not have data on all
4 years of the employee satisfaction measures. As such, the
primary analysis uses a full information maximum likelihood
imputation method for SEM to impute all missing data for the
employee data. This method produces more reliable results for
longitudinal data than listwise deletion, even when 75% of the
data is not missing at random (Newman 2003).
2
To begin the analysis, we modeled different forms of latent
growth for employee satisfaction because misspecification of
a latent growth can lead to inaccurate estimates of that
growth’s influence on dependent variables (Bollen and
Curran 2006; Steenkamp and Baumgartner 2000). Per latent
growth modeling convention (Bollen and Curran 2006), the
loadings for the intercept were fixed to 1, whereas the factor
loadings for the linear trajectory were fixed in a sequential
order of 0, 1, 2, and 3. A linear growth model provided a
stronger fit to the data than an intercept-only model.
Furthermore, the intercept (ψ
αα
=.11, p< .001) and slope
(ψ
β1β1
=.01,p< .01) variables of the linear slope model ex-
hibited statistically significant variance, indicating that the
companies in the data set had varying levels and trajectories
of employee satisfaction. A quadratic growth model only mar-
ginally improved model fit (Δlog-likelihood: 8.34, Δdf = 4,
p= .08). To maintain parsimony, we utilized the simpler linear
growth model. Test statistics and coefficients for the growth
models are available in Web Appendix C.
To ensure omitted variables did not bias the results, we
included several control variables. Company size can affect
both employee and customer satisfaction, possibly through
market share (Anderson et al. 1994;Beer1964;Hogreve
et al. 2017). We control for size by including a variable based
on the number of employees that filled out a Glassdoor review
in 2014.
3
In addition, recent research suggests company
downsizing and financial losses can simultaneously affect em-
ployees and customers (Habel and Klarmann 2015). In accor-
dance with Habel and Klarmann (2015), we included a series
of dichotomous variables to account for the presence of a
downsizing and financial loss in 2014. We also included var-
iables representing the number of times a firm experienced
downsizing or financial loss in the last 4 years. Finally, we
included a series of industry-based dummy variables to ac-
count for heterogeneity in customer satisfaction across indus-
tries that may confound any observed effects of employee–
customer contact. The equations representing the overall anal-
ysis are below:
EMPSATit ¼τiþλtπ1i þρit ð1Þ
CUSTSATi¼αþβ1τiþβ2π1i þβ3CONTACT
þβ4τix CONTACT þβ5π1i
x CONTACT þβ6−24CONTROLiþεi;
ð2Þ
where EMPSAT is the average employee satisfaction for com-
pany i at time t, τis the initial level of employee satisfaction
for company i, π
1
is the linear slope in employee satisfaction
for company i, λ
t
is the set loading for time t, and CUSTSAT is
the average customer satisfaction for company i. CONTACT
is a dichotomous variable coded as zero or one,in which a one
represents customers having medium to high contact with em-
ployees as a normal part of consumption. CONTROL is a
series of control variables for company i. The interactions
between the growth curve latent variables and the dichoto-
mous contact variable were created and estimated using the
Latent Moderated Structural Equations technique in Mplus
because such an approach is robust to multicollinearity and
good at producing unbiased estimates (Kelava et al. 2008). All
independent variables were mean centered before the analysis
to ease interpretation of the interaction coefficients.
Results
Preliminary analyses Though our industry controls and lagged
dependent variable help reduce some concerns of
endogeneity, such concerns still exist. We took two additional
steps to further address concerns of endogeneity. First, in a
subsequent analysis, we controlled for earlier lagged versions
of our dependent variable to control for potentially stable and
exogenous effects related to customer satisfaction. Second, we
conducted a Hausmann test using instrumental variables.
Following the procedure of Mani et al. (2014), we used lagged
versions of our independent variables under the assumption
that such instruments would extract the stable and exogenous
factors of employee satisfaction level and trajectory. This
2
We still find support for the hypothesis when restricting the sample to only
the companies with complete data (n= 262), but the results are slightly weaker
than those reported in the primary analysis (β
4
= 6.88, t = 2.64, p<.01;β
5
=
33.12, t = 2.79, p< .01).
3
We compared the survey-based measure of company size to one basedon the
number of employees for publicly traded companies. The correlation was
strong (r = .39) considering modern analyses of correlational effect sizes
(Bosco et al. 2015). Furthermore, an analysis using the employee-based mea-
sure found similar results and still supported the hypotheses despite a smaller
sample size (n=202).
J. of the Acad. Mark. Sci.
analysis, reported in Web Appendix D, revealed that our in-
dependent variables exhibited sufficient freedom from
endogeneity bias.
Hypothesis testing (H1–H3) We now turn the focus to hypoth-
esis testing (see Table 3). The direct effect of employee satis-
faction level was marginally significant (β
1
= 2.05, t = 1.82,
p= .07) whereas the direct effect of employee satisfaction tra-
jectory was insignificant (β
2
= 1.78, t = 0.32, p= .75). In con-
trast, both the interaction of contact with employee satisfac-
tion level was positive and significant (β
4
= 8.55, t = 4.24,
p< .01) as was the interaction of contact with employee satis-
faction trajectory (β
5
=40.57,t=3.93,p< .01). Thus, the ef-
fect of both the level and trajectory of employee satisfaction
on customer satisfaction were enhanced by high employee–
customer contact, which provides support for H1 and H2.
To test H3, we first partitioned the low level of the
employee–customer contact variable into two separate catego-
ries: those firms with no employee–customer contact and
those with minimal contact as a normal part of consumption.
The resulting contact variable contained three categories (i.e.,
no, minimal, and high contact). We then coded the contact
variable using contrast codes. Though the contact variable
reflects increasing and decreasing levels of contact, it is an
ordinal variable. Using such a variable in empirical tests for
continuous variables would result in inaccurate coefficients.
The first contrast (labeled C1) coded the lowest level of con-
tact as −.66, whereas the other two levels were coded as .33.
This coding scheme assesses the significance of the difference
between no employee–customer contact and any form of con-
tact. The second contrast (labeled C2) coded the no contact as
zero, low contact as −.50, and high contact as .50. This coding
scheme partitions the variance uniquely due to the difference
in contact between the low and high levels from the other
contrast variable. We then created interaction terms with the
level and trajectory of employee satisfaction for both categor-
ical variables, resulting in a total of four interaction terms.
To avoid convergence problems in Mplus due a large num-
ber of latent interactions, we tested H3 using ordinary least
squares (OLS) regression. The scores for each company’s
2011 2012 2013 2014
2.0
3.0
4.0
Samsung
Jones Group
Northwestern Mututal
H-E-B
Proctor & Gamble
PPL
Lile Caesar’s
Delta
eBay
AAA
2.5
3.5
noitcafsitaseeyolpmE
Fig. 3 Employee satisfaction for
ten randomly chosen companies
from 2011 to 2014 based on data
from Glassdoor.com.Notes:AAA
American Automobile
Association, PPL PPL
Corporation, H-E-B HEB
Grocery Company, LP
Table 3 Study 1 results: The effect of employee satisfaction trajectory
on customer satisfaction
Customer satisfaction
−2 log-likelihood 1284.61
AIC 2639.22
BIC 2656.07
P Coefficients Standard errors
Intercept α69.83 (3.64)***
Employee satisfaction level β
1
2.05 (1.12)+
Employee satisfaction trajectory β
2
1.78 (5.63)
Contact β
3
−17.19 (5.87)**
Satisfaction level X contact β
4
8.55 (2.02)***
Satisfaction trajectory X contact β
5
40.57 (10.32)***
Controls
ln (Company size) β
6
0.77 (0.48)
ln (Company relative size) β
7
−1.04 (0.52)*
ln (Number of competitors) β
8
−0.39 (0.76)
Downsizing β
9
−0.44 (1.20)
Total prior downsizings β
10
0.19 (0.44)
Financial loss β
11
−0.38 (2.22)
Total prior financial losses β
12
−0.21 (0.70)
Automobile manufacturers β
13
4.02 (1.12)***
Banks β
14
−7.47 (2.27)**
Clothing manufacturers β
15
2.38 (2.05)
Restaurants β
16
−4.17 (1.88)*
Food manufacturers β
17
3.32 (1.40)*
Good manufacturers β
18
0.77 (1.76)
Insurance providers β
19
−0.86 (1.52)
Phone service providers β
20
−10.06 (1.84)
Retailers β
21
−5.86 (1.74)*
Shipping companies β
22
−1.72 (3.43)
Travel companies β
23
−9.84 (1.96)***
Internet-based companies β
24
−0.69 (1.29)
+p<.1.*p<.05.**p<.01.***p<.001
Numbers in parentheses are standard errors. Numbers to the left of the
parentheses are the non-standardized coefficients. Pparameters. Contact
is a dichotomous variable in which a zero (one) represents a company’s
employees and customers have little-to-no (medium-to-high) interaction
as a part of consumption. Company size is the number of employees who
filled out a Glassdoor.com survey in 2014. Company relative size is the
company size variable divided by the number of employees who filled out
aGlassdoor.com review in that company’s industry. The industry
variables are dummy variables compared against utility companies
J. of the Acad. Mark. Sci.
employee satisfaction level and trajectory were estimated
using the same latent growth model as before but were then
exported into the regression analysis. The results of this anal-
ysis revealed the interaction of C1 and the level of employee
satisfaction was significant (β= 8.30, t = 3.96, p<.01),
whereas the interaction of C2 and the level of employee sat-
isfaction was insignificant (β= 2.43, t = 0.74, p=.46).Thus,
H3a was supported because the transition from no contact to
any contact resulted in a significant increase in the direct effect
of employee satisfaction level on customer satisfaction, but
greater levels of contact resulted in no other significant in-
crease. In contrast, the interaction of employee satisfaction
trajectory with C1 was insignificant (β= 17.07, t = 1.53,
p= .13) whereas the interaction with C2 was significant
(β=44.17,t=2.39,p= .02). Thus, H3b is supported because
high contact facilitated the effect of the employee satisfaction
trajectory, whereas lower degrees of contact did not. The di-
rect effects, as shown in the bottom of Fig. 2, further illustrate
that the effect of the trajectory is only significant at high con-
tact whereas the effect of level became significant with low
contact.
Are the observed effects robust to the stability of customer
attitudes? As the research in Table 1suggests, employee
satisfaction may not affect customer satisfaction when the
effect of customer satisfaction over time is controlled
(Zablah et al. 2016). To account for this possibility, we
included the previous year’s customer satisfaction (i.e.,
the ACSI data for 2014) in the primary analysis (used to
test H1 and H2) as a robustness check. The ACSI did not
measure several companies across both years (a total of
223 unique firms matched across the 2 years of our ACSI
data) resulting in the loss of nearly 25% of the data. As
expected, the effect of the previous year’s customer satis-
faction was very strong (β= 0.71, t = 16.93, p< .001).
Despite this control and sample size reduction, the contact
interaction with employee satisfaction trajectory was sig-
nificant (β
5
= 12.71, t = 1.98, p= .048), though the inter-
action of contact with employee satisfaction level became
marginally significant (β
4
= 2.34, t = 1.77, p= .08). The
direct effect of employee satisfaction level (β
1
=0.33,
t = 0.46, p= .65) and employee satisfaction trajectory
(β
2
=−0.59, t = −0.17, p= .87) remained insignificant.
Thus, the proposed effect of employee satisfaction trajec-
tory affecting customer satisfaction through employee–
customer contact is robust to controlling for customer sat-
isfaction over time.
Are the observed effects robust to the last measure of em-
ployee satisfaction? A critique of the previous analysis is that
comparing the effect of a satisfaction trajectory against an
initial satisfaction level (as is done in typical latent growth
modeling) unfairly biases the analysis in favor of the
trajectory. Stated differently, it could be that the trajectory
effect for companies with employee–customer contact is a
result of the last employee satisfaction measure rather than
the trajectory from the initial satisfaction measure. To account
for this possibility, we re-specified the latent growth model so
that the trajectory was compared against the last measure of
employee satisfaction rather than the first. More specifically,
we changed the loadings for the latent growth factor to be
ascending and negative (−3, −2, −1, 0) so that the intercept
is the last measure (Bollen and Curran 2006). Using negative
values with ascending numbers (rather than specifyinggrowth
loadings as 3, 2, 1, 0) ensures that positive numbers for the
slope coefficients represent increasing employee satisfaction.
When the employee satisfaction level is based on the last
measure of employee satisfaction, the employee–customer
contact interaction for employee satisfaction trajectory visibly
weakened but remained marginally significant (β
5
= 21.19,
t = 12.00, p= .08). Employee satisfaction level’s direct effect
(β
1
= 2.02, t = 1.73, p= .08) and interaction with employee–
customer contact (β
4
= 7.06, t = 3.19, p< .01) remained virtu-
ally unchanged. Thus, the moderated effect of employee sat-
isfaction by employee customer contact was robust to the
choice of the employee satisfaction measure used to represent
the employee satisfaction level.
Discussion
The results in Study 1 provide strong evidence that changes in
employee satisfaction affect customer satisfaction and that
employee–customer contact facilitates this effect. Without
employee–customer contact, the effect of the trajectory is non-
existent. Importantly, the current results fail to eliminate the
possibility that changes in customer satisfaction (i.e., a cus-
tomer satisfaction trajectory) may be a better predictor of
changes in employee satisfaction. We attempted to rule out
this possibility, but the growth model on ACSI scores concur-
rent with the Glassdoor data (from 2011 to 2014, n=176)was
not positive definite and the variance of the slope was nega-
tive. In other words, a latent growth model did not fit on a time
series of customer satisfaction. However, Study 1 does dem-
onstrate that employee satisfaction trajectories are predictive
of customer satisfaction, even when taking into account prior
customer satisfaction, current employee satisfaction, and firm
characteristics.
Study 2
Data structure
To provide another test of the proposed framework, we
accessed data from a national rental car company based in
J. of the Acad. Mark. Sci.
the U.S. made available by the Wharton Customer Analytics
Initiative. This data includes surveys from employees and
customers along with transaction data. Our analyses only in-
cluded employees who worked at the rental locations rather
than other locations (e.g., headquarters, collections). Both the
customer and employee surveys were anonymous, but were
nonetheless linked through the retail store where the rental
vehicle was acquired and where the employee was located.
The company linked employees to a location only when sur-
vey participation was high enough to uphold anonymity (n>
3). As such, only 102 locations were identifiable, of which
only 91 contained full data for all of the utilized employee
surveys. The employees were measured across four waves that
spanned 20 months and the number of employees participat-
ing in the surveys was similar across the time periods (wave
1 = 7574; wave 2 = 8150; wave 3 = 7451; wave 4 = 7696),
which provided a total of 30,871 employee responses. An
average of 45 employees per store participated in each survey,
with a range from 4 to 189.
The customer data were from surveys offered at the time a
transaction was finished. The specific customer data used in
the analyses were collected during the month of the fourth
employee survey and the following 3 months. This observa-
tion window provided 43,305 customer surveys linked to the
91 rental centers. However, our contact and control variables
required using customers in the loyalty club (as explained in
more detail later), which reduced the final sample size to
36,208 customers. We used the same non-response bias anal-
ysis as used in Study 1 and once again found little practical
significance in the results (which are reported in Web
Appendix B).
Variable operationalization
The employee data consisted of nine items that resemble those
used for perceived organizational support, job satisfaction,
and employee engagement. A subsequent study, reported in
Web Appendix E, demonstrated a lack of discriminant validity
between five of the nine items and typical measures of em-
ployee satisfaction (e.g., BAll in all, I am very satisfied with
my current job^). Therefore, these five items served asa proxy
for employee satisfaction (see Appendix Table 6).
4
The cus-
tomer data did not measure satisfaction, but instead measured
customers’repatronage intentions with a single item.
Importantly, repatronage intentions are theoretically relevant
because repatronage intentions result from customer satisfac-
tion (Heskett et al. 1997).
We operationalized varying degrees of contact by tracking
the number of times customers had frequented the rental
center in which a customer survey was completed, in accord
with previous research (Dietz et al. 2004). The rental car com-
pany maintains a loyalty program and a large majority of the
surveys within the observation window for the rental centers
were completed by customers in the program (83%).
Transaction data was linked to both the loyalty club accounts
and the store location of the survey, which allowed us to track
the number of times a loyalty club member rented a vehicle
over the 2 years that employee satisfaction was measured.
Any customer who was not part of the loyalty club was re-
moved because we could only speculate on their patronage.
Because the patronage frequency variable was skewed, the
natural log of the variable was used. All measures are shown
in Appendix Table 6and more measurement statistics, includ-
ing a correlation matrix, are shown in Appendix Table 7.
Analytical technique
Theemployeedataisanonymous.Assuch,thedatamustbe
aggregated to the store level (i.e., level 2) to be able to link the
four measurements together. Reliability and agreement statis-
tics on the satisfaction measure within each survey justify such
aggregation (ICC1 = .10–.14, ICC2 = .83–.87., Lebreton and
Senter 2008). Once the data was aggregated to the store level,
we assessed the within- and between-group variance in em-
ployee satisfaction over time with an ICC statistic, as done in
Study 1. Once again, a majority of the variance in employee
satisfaction over time was attributable to within-group vari-
ance (55%), whereas a minority was attributable to between-
group variance (45%). This finding further supports our focus
on employee satisfaction trajectories as a representative metric
of the within-group variance.
To capitalize on the full size and variance of the customer
data, a multilevel model was used because it provides better
estimates of coefficients and standard errors than aggregating
or disaggregating the data (Raudenbush and Bryk 2002). The
customer data was kept disaggregated at level 1 such that the
latent growth variables were used to predict the randomly
varying intercept of repatronage intentions. The 91 stores for
the level 2 grouping structure is far above the sample size
threshold of 50 at which reliable estimates are typically pro-
duced in multilevel models (Maas and Hox 2005).
We first modeled the employee satisfaction trajectories to
establish the correct form for the data by transitioning through
a series of models similar to Study 1. A linear growth model
proved to be the most parsimonious fit to the data once again
(Δlog-likelihood: 19.1, Δdf = 3, p< .01) because a quadratic
growth model did not result in a significant improvement in
model fit (Δlog-likelihood: 5.87, Δdf = 4, p= .21; See Web
Appendix F). To test the effect of employee satisfaction tra-
jectories on the customer variables, we specified the latent
growth factors at level 2 as affecting the randomly varying
4
We also tested the single recommendation item given it is most representa-
tive of employee satisfaction as a robustness check. We found the coefficients
related to hypothesis testing strengthened when only this item was used.
J. of the Acad. Mark. Sci.
intercept of repatronage intentions and the slope of the effect
of the contact variable on repatronage intentions.
We included several controls that were available in the
dataset. Repatronage and employee satisfaction may vary
as a function of the type of customer; as such, we con-
trolled for whether a customer was a business (versus
non-business) customer. Furthermore, employee and cus-
tomer affective states can be regionally determined, so we
controlled for the region of the rental center a customer
patronized (Govind et al. 2018). We controlled for rental
center size based on the number of employees that com-
pleted the employee survey and growth of the rental cen-
ter by subtracting the number of employees that filled out
the first survey from the number that filled out the last
survey. We controlled the average tenure of employees at
the rental center by averaging employees’responses to the
survey question of how long they had worked with the
company). We further controlled for the value of the rent-
al based on the customer survey data and the cost of the
rental based on the transaction data. The level 2 controls
were specified as affecting the level 1 intercepts as shown
in the following equations:
EMPSATjt ¼αjþλtβjþrjt;ð3Þ
REPATij ¼π0j þπ1j CONTACTij þπ2j VALUEij
þπ3j COSTij þπ4j PURPOSEij þεij;ð4Þ
π0j ¼γ00 þγ01 αjþγ02 βjþγ03 SIZEj
þγ04GROWTHjþγ05 TENUREjþγ06−10
REGIONjþμ0j ;
ð5Þ
π1j ¼γ10 þγ11 αjþγ12 βjþμ1j;ð6Þ
EMPSAT is rental center j’s average employee satisfaction
at time t, αis the initial level of employee satisfaction for
rental center j, βis the linear slope in employee satisfaction
for rental center j, λ
t
is the set time-based loading for time t.
REPAT is the ith customer’s stated repatronage intentions in
relation to rental center j and CONTACT is the ith customer’s
number of transactions with rental center j up to their survey
response. VALUE, COST, and PURPOSE are the ith cus-
tomer’s evaluations of rental value, rental cost, and rental pur-
pose in relation to his/her experience at rental center j, respec-
tively. SIZE, GROWTH, and TENURE are rental center j’s
number of employees, growth in employees, and average em-
ployee tenure, respectively. REGION represents a series of
dummy variables denoting the region of rental center j.
Finally, μare random effects in the intercept and slope asso-
ciated with the jth rental center not explained by the included
variables. All independent variables were centered at their
grand mean to facilitate interpreting the coefficients
(Raudenbush and Bryk 2002).
Results
Hypothesis testing As shown in Table 4, the coefficient for the
interaction between employee satisfaction level and contact
was insignificant (γ
11
=0.01, t=0.57, p= .57). As such, the
data did not support H1. Furthermore, the direct effect of the
initial employee satisfaction level was significant (γ
01
=.08,
t=2.02,p= .04). Thus, employee satisfaction level exhibited
a positive relationship with repatronage intentions regardless
of whether a customer’s contact with employees was frequent
or limited to a single interaction. In contrast, the coefficient for
the direct effect of employee satisfaction trajectory was insig-
nificant (γ
02
=0.59,t=1.39,p= .17) whereas the interaction
with contact was marginally significant (γ
12
=0.17,t=1.88,
p= .06). Thus, the data supported the hypothesis (H2) that
repeated contact increases the effect of employee satisfaction
trajectory on customers’repatronage intentions.
To test H3, we broke apart the effect of low and high con-
tact. To do so, we created a new variable by first mean cen-
tering the contact variable and then substituting zero for every
value below the mean. We then included this variable into the
analysis in addition to the original contact variable. In this
way, the new variable only examines the unique moderating
effect attributable to high contact.
5
When the new variable
was included, the interaction of employee satisfaction trajec-
tory with contact became insignificant (γ=−0.15, t=−1.17,
p= .24). In contrast, the new variable that only accounted for
the moderating effect of high contact was strong and signifi-
cant (γ=0.27,t=−2.20, p= .03). Thus, the analysis provided
support for H3b because the effect of employee satisfaction
trajectory on customer satisfaction was dependent on high
levels of employee–customer contact. The direct effects of
employee satisfaction trajectory at different levels of contact,
shown in the bottom of Fig. 2, further illustrate the significant
moderating effect of high contact. The interaction of employee
satisfaction level with either contact variable was insignifi-
cant. Thus, the data did not support H3a.
Do employee satisfaction trajectories affect future revenue?
To test whether employee satisfaction trajectories affected sales,
we calculated a future revenue variable for each rental center
basedonthe3monthsoftransactionsattheendofthe2year
5
A similar way to test the unique effect of high contact is to create a three-way
interaction between employee satisfaction trajectory, contact, and a dummy
variable that has zeros for values of contact below the mean and a one for
values above the mean. Doing this revealed that the three-way interaction was
positive and significant (γ=2.68.,t=2.78,p= .01). The two-way interaction
between contact and trajectory was insignificant (γ=−0.82., t = −1.26,
p= .21). Yet another way to test H3b using the continuous patronage variable
is by a curvilinear moderation through a quadratic interaction in which a
squaredformofthemoderatorisusedinadditiontothenon-squaredform.
Doing this further supports the necessity of high contact for employee satis-
faction trajectory as the quadratic interaction was significant and positive (γ=
0.27, t = 2.20, p= .03) whereas the linear interaction was negative (γ=−0.18.,
t=−1.21, p=.23).
J. of the Acad. Mark. Sci.
observation window. To control for the number of customers,
we counted the customers for each rental center during this
same period and divided the revenue variable by the number
of customers. Doing this transformation helped normalize the
future revenue variable and kept the coefficients in terms of
dollars.
6
The financial metric variable, at level 2, was predicted
by repatronage intentions at level 1. This kind of specification
in multilevel modeling is not common, but is legitimate (Croon
and Van Veldhoven 2007). In essence, Mplus utilizes the vari-
ance of the level 1 variables that can be attributed to the rental
centers to predict the financial dependent variable. To assess the
mediated effect of employee satisfaction trajectory at high
employee–customer contact, we utilized the customer data
above the median of the repeat variable (i.e., two visits to a
rental center). We also employed the bootstrapping procedure
in Mplus (based on 5000 bootstrapped samples) to estimate
standard errors and confidence intervals for the indirect effect
(Zhao et al. 2010). The control variables at level 1 (value per-
ceptions, rental cost, and rental purpose) were also allowed to
affect the level 2 dependent variable to ensure the effects of
repatronage were unique. The results revealed a positive effect
on future revenue for repatronage intentions (γ=63.54, t=
2.76, p< .01). In addition, the mediated effect of employee
satisfaction trajectory through repatronage intentions was mar-
ginally significant (a x b = 42.75, 90% CI: 1.37 / 84.14). Thus,
employee satisfaction trajectories exhibited an indirect relation-
ship with future revenue through increased customer loyalty, as
indicatedbyrepatronage intentions.
Are the observed effects robust to the stability of customer
attitudes? As done in Study 1, we sought to control for the
persistence of customer attitudes over time by including a
6
A robustness check using the future revenue variableby itself and controlling
for total customers did not result in changes to the direction and significance of
the estimated coefficients. This analysis required a natural logarithmic trans-
formation of the revenue and customer variables to offset their skew.
Table 4 Study 2 results: The
effect of employee satisfaction
trajectory on customer outcomes
across customer groups with
differing amounts of employee
contact
Repatronage intentions
2 Log-Likelihood 200,796.04
AIC 401,788.08
BIC 402,620.79
P Coefficients Standard errors
Intercept γ
00
3.29 (0.16)***
Employee satisfaction level γ
01
0.08 (0.04)*
Employee satisfaction trajectory γ
02
0.59 (0.43)
ln (Contact) γ
10
0.02 (0.07)
ln (Contact) X Employee satisfaction level γ
11
0.01 (0.02)
ln (Contact) X Employee satisfaction trajectory γ
12
0.17 (0.09)+
Value perceptions π
2
0.40 (0.00)***
ln (Rental cost) π
3
0.04 (0.00)***
Purpose (0 = leisure, 1 = business) π
4
0.02 (0.01)*
ln (Size of rental center) γ
03
0.03 (0.02)+
Growth of rental center γ
04
0.00 (0.00)
Average employee tenure γ
05
0.05 (0.04)
Western region γ
06
0.07 (0.04)+
Southeastern region γ
07
−0.00 (0.05)
Northeastern region γ
08
0.02 (0.04)
Southwestern region γ
09
0.00 (0.05)
Central region γ
10
0.01 (0.06)
+p<.1.*p<.05.**p<.01.***p<.001
Numbers shown are unstandardized coefficients with standard errors in parentheses. PParameters. Contact is a
repatronage variable representing the amountof a customer’s patronage with a specific rental center over the two-
years observation window. Size of location is the average number of employees who filled out a survey at a rental
center (across the four surveys). The regional variables are dummy variables compared against rental centers
located outside the United States
J. of the Acad. Mark. Sci.
previous measure of repatronage intentions. We used the
customer survey data again to isolate customer data for
each rental center at the time of the last employee survey.
These data were aggregated to the rental center, grand
mean centered, and included in the same analysis used
to test H1 and H2 as level 2 variables affecting the inter-
cept of repatronage intentions. As expected, the effect of
previous repatronage intentions on current repatronage in-
tentions was significant (γ=0.26, t= 2.94, p< .01). The
coefficient for the interaction between employee–
customer contact and employee satisfaction trajectory
remained marginally significant (γ=0.38, t= 1.84,
p= .07) though the direct effect of employee satisfaction
leveldidnot(γ=0.06, t= 0.96, p= .34). Thus, the effect
of trajectory, as enabled by high employee–customer con-
tact, was robust to the stability of customer attitudes over
time whereas the effect of employee satisfaction level was
not.
Are the observed effects robust to the last measure of em-
ployee satisfaction? We replicated the competing model
test from Study 1 by re-specifying the loadings of the
growth model in an ascending, negative order (−3, −2,
−1, 0) so that the intercept, and thereby employee satis-
faction level, was based on the last measure of employee
satisfaction rather than the first. With the new loadings,
the interaction between employee–customer contact and
employee satisfaction trajectory just barely remained mar-
ginally significant (γ=0.21,t= 1.65, p=.10) whereas the
direct effect of the level remained virtually unchanged
(γ=0.09,t=2.06,p= .04). Thus, the data once again sup-
ports a latent growth trajectory in employee satisfaction
affecting customer satisfaction (when employee–customer
contact is high) even when the latest employee satisfac-
tion scores are controlled.
Discussion
Study 2 extended Study 1’s observed effects of employ-
ee satisfaction trajectories to a narrower timeframe, new
customer metrics, and a more precise measure of contact
at the customer level (repeat patronage behavior). The
interaction between employee satisfaction trajectory and
employee–customer contact was supported again.
Controlling for previous customer metrics did not dis-
rupt this relationship, nor did controlling for recent em-
ployee metrics. A key difference across the two studies
was that the employee satisfaction level did not exhibit
an interaction with contact in Study 2, but it did in
Study 1. This difference may be because Study 1 ex-
amined firms with no employee–customer contact,
whereas all customers had at least one point of contact
in Study 2. In other words, the effect in Study 1 may
have been driven by the difference between no contact
and some contact. If so, this possibility still supports
our proposed theory, illustrated in Fig. 2, which sug-
gests the interaction of employee satisfaction level and
employee–customer contact on customer metrics is most
apparent when contact ranges between non-existent and
low levels. In contrast, the interaction between employ-
ee satisfaction trajectory and employee–customer contact
on customer metrics is most apparent at high levels of
contact.
General discussion
The two studies currently reported provide strong evidence
that employee satisfaction exhibits growth patterns over time
(i.e., trajectories) that affect customer satisfaction and
repatronage intentions. The degree of customer-employee in-
teraction moderated the effect of trajectories across both
studies―so much so that trajectories had no significant effect
for companies with low employee–customer contact in Study
1 or for customers who had minimal patronage in Study 2.
Notably, across both studies, the effect of the employee satis-
faction level on customer metrics maintained its strength even
in instances of less employee–customer contact. Collectively,
these findings yield key contributions that offer theoretical
and managerial insight.
Theoretical contributions
The research listed in Table 1suggests that changes in
employee satisfaction do not affect customer satisfaction.
Our results offer a counter explanation. The current re-
sults, based on different levels of employee–customer
contact, provide strong evidence that customers’satisfac-
tion and repatronage can be influenced by improving em-
ployee satisfaction. Although previous research has
established the importance of moving beyond static ap-
proaches by incorporating the dynamic nature of con-
structs, such as employee satisfaction (Chen et al. 2011;
Liu et al. 2012), customer commitment (Palmatier et al.
2013), and customer-company identification (Huang and
Cheng 2016), no work to date examines whether changes
on one side of the organizational frontline can affect the
other side. Research has examined how the actions of
employees affect customers’relationship velocity
(Harmeling et al. 2015; Palmatier et al. 2013); however,
this is not the same as examining how dynamics in the
employee-organization relationship affect the customer-
organization relationship. Our research remedies this the-
oretical omission from prior research by providing this
link. Customers, at some level, are perceptive to system-
atic changes in frontline employees’satisfaction. Not only
J. of the Acad. Mark. Sci.
are customers perceptive of the change in employees’dis-
position, but customers also exhibit strong affective (i.e.,
satisfaction) and conative (i.e., repatronage intentions) re-
actions to this change.
Further, our findings identify high employee–customer
contact as the mechanism by which trends in employees’
satisfaction spill over to customers. We demonstrate that
contact is imperative for employee satisfaction trajectories
as contact fosters situational involvement, permitting the
deeper, inferential processing of employee behavior.
Repeated contact forces customers to be more cognizant
of discrepancies in employee states and attentive to cues
that provide evidence of employee trajectories. By serving
as disconfirming evidence, these cues then inform cus-
tomers’service evaluations and satisfaction. We also pro-
vide a clarification of previous research (as shown in
Tab le 2) that has shown mixed results of the moderation
of employee satisfaction level by employee–customer
contact. By showing that this moderation effect is most
pronounced at lower levels of contact, which facilitate
affective reactions processes, the current research suggests
the mixed results were at least partly dependent on the
level of contact that was assessed by past research.
The distinguishable effect of employee satisfaction trajec-
tories from levels and their different interactions with
employee–customer contact exemplifies the recent recom-
mendation by Certo et al. (2016) for researchers to use longi-
tudinal research to separate within- and between-group vari-
ances in relation to well-studied phenomenon. Employee sat-
isfaction trajectories captured within-firm and within-unit var-
iance into a single metric that allowed us to isolate its unique
effects and properties from that of levels. This separation fa-
cilitated the observation of the interaction of employee satis-
faction level with low employee–customer contact in Study 1.
In a posthoc analysis, we removed the effect of the trajectory,
which resulted in the level exhibiting an interaction with both
high and low contact when testing H3 (rather than just low
contact as shown in the results).
The connection of the current research to the larger body
of research on relationship dynamics creates several direc-
tions for future areas of inquiry. One next step would be
linking changes in employees to changes in customer satis-
faction, customers’relational states, or a transformational
relationship event (Harmeling et al. 2015). For example, it
would be interesting to know how changes in customers,
such as those represented in relationship velocities or satis-
faction trajectories, affect changes in employees. Does con-
tact facilitate and increase this outside-in effect? Further,
how do customer defections affect employee satisfaction
trajectories? In fact, the limited research on employee satis-
faction trajectories has focused on outcomes rather than an-
tecedents (Chen et al. 2011; Liu et al. 2012), producing the
need for research to establish the drivers of employee
satisfaction trajectories and whether such drivers differ
from those of the static level. This approach may require
tracking new employees to observe a true beginning state.
In addition, modeling trajectories simultaneously at differ-
ent vertical levels, such as the trajectory of companies, sub-
units, and employees, would provide an opportunity to ob-
serve how trajectory alignments or misalignments affect
customer judgments.
Future work could also evaluate the extent to which exter-
nal forces reinforce trajectories, or whether the trajectories fail
to need reinforcements and propel upward or downward on
their own. Such a focus would be meaningful because firms
may or may not need to deploy resources to continue or ob-
struct a spiral. Similarly, research might also assess the signif-
icance of corrected trajectories. Research could address ques-
tions such as: what are the downstream effects of negative
trajectories that have turned to become positive trajectories?
Can corrected trajectories positively influence customer out-
comes, and if so, how does their effect compare to trajectories
that have been consistently improving? Consistent with the
need for future research to expand the model we test, exam-
ining the cause of trajectory change and additional down-
stream effects would advance knowledge by creating an im-
portant theoretical nuance to our understanding ofrelationship
dynamics.
Managerial contributions
Companies spend nearly $720 million annually to im-
prove employee satisfaction (Kowske 2012), so questions
remain regarding whether such substantial investments are
justified. Our research shows that investments in employ-
ee satisfaction are worthwhile because they extend be-
yond the firm to affect customer satisfaction and
repatronage intentions. Given our findings, we can answer
the hypothetical question asked in the introduction about
which employee would more favorably affect customer
judgments. Is Employee B (i.e., higher satisfaction level
but negative trajectory) more impactful than Employee A
(i.e., lower satisfaction but positive trajectory)?
To answer this question, we utilize the results of Study 2.
Given our analyses were at the business unit level, we
equate the two employees with different rental centers. An
improving trajectory such as Employee A’s is approximate-
ly one standard deviation above the mean, whereas the level
is approximately three standard deviations below the mean.
Employee B’s trajectory is approximately three standard
deviations below the mean and the level is approximately
one standard deviation above the mean. For customers with
contact above the mean (i.e., more than two visits over
2 years), Employee A results in a repatronage score of
3.72 (out of five) and Employee B results in a repatronage
score of 3.59 (holding all controls at their mean). Thus,
J. of the Acad. Mark. Sci.
based on the data in Study 2, Employee A actually results in
a greater impact on customer evaluations than Employee B.
Given the reported effect of repatronage intentions on future
revenue, the trajectory of Employee A’s satisfaction would
result in an additional $18.23 per customer (on average)
over 3 months as compared to Employee B’s satisfaction.
Given there were approximately 13,000 customers over the
three-months period tied to the 89 rental centers used in our
analyses, that difference in revenue per customer equates to
$237,000 in additional total revenue if customers are served
by employees more like Employee A than B.
Notably, investments in improving employee satisfac-
tion with a purpose of improving customer satisfaction as
outlined above are conditional on the extent to which a
firm’s customers have multiple encounters with em-
ployees. For companies with little employee–customer
contact or few repeat customers, we find effects for the
level of employee satisfaction, but not for changes in sat-
isfaction. Thus, managers of low employee–customer con-
tact firms benefit from considering improving employee
satisfaction as a secondary means of influencing customer
metrics, whereas managers at high employee–customer
contact firms benefit from considering improving employ-
ee satisfaction as one of the primary means to influence
customer outcomes. As such, managers at high employee–
customer contact firms need to take a longitudinal per-
spective in the consideration of and reward for employee
satisfaction. The last measurement of employee satisfac-
tion or satisfaction across business units should be com-
bined with several measurements from the past to create a
full picture of where the business units stand.
As a last consideration, firms with little employee–
customer contact may want to entertain ways in which to
facilitate such contact to the extent the firm wants to leverage
the benefits of improving employee satisfaction. Similar to the
idea that embedding customers within the fabric of the orga-
nization facilitates deep customer-company relationships
(Bhattacharya and Sen 2003), such embedding allows for
thetrajectoryofemployeesatisfactiontodeepenthe
customer-firm relationship by influencing customer out-
comes. The methods for enabling contact range from opening
retail outlets that bring the organization to the customer (e.g.,
Apple stores) to recruiting customers as volunteers for orga-
nizational processes in order to bring the customer to the or-
ganization (e.g., Southwest Airlines’hiring process). In sum-
mary, organizations may want to remove barriers to
employee–customer contact, including barriers such as dis-
tance or technology (Giebelhausen et al. 2014).
Limitations
Admittedly, our work is not without limitations. One short-
coming includes our limited inclusion of mediators and
moderators. One possible avenue for future research is to
examine the length of time a customer has interacted with an
employee (i.e., relationship length). Compared to new rela-
tionships, established relationships are often characterized
by a stronger dependence on frontline employees because
these employees are key to value creation (Sirdeshmukh
et al. 2002;Zablahetal.2016). Another limitation of our
study is the possibility of endogeneity. In reality, no re-
search can ever rule out the possibility of endogeneity, but
we took what steps were possible to assess the likelihood of
bias from endogeneity. Nonetheless, it would be useful for
future research to employ robust endogeneity controls to
better assess population values of the effect of employee
satisfaction trajectories on customer outcomes.
An important limitation is that the current study only
assessed effects in one direction, the effect of employee
satisfaction on customer outcomes. Research suggests that
this effect may not be unidirectional, and instead supports
outside-in effects, wherein customer satisfaction influ-
ences employee satisfaction (Zablah et al. 2016). The cy-
clical and amplifying nature of these effects points to the
importance of continued research in this area. That is,
decreases in employee satisfaction likely lead to decreases
in customer satisfaction, which in turn lead to further de-
creases in employee satisfaction, and so on. Most likely,
the effects ebb and flow between the two directions.
Given this possibility, more longitudinal research is need-
ed to track employee and customer satisfaction over a
prolonged period with many measurement points to ascer-
tain when equilibrium is reached and the predominant
direction of effects. Likely, this form of research will need
to capitalize on advanced dynamic modeling that can in-
corporate many time measurements to uncover new in-
sights (Asparouhov et al. 2018).
Finally, our work did not investigate employee–
customer contact occurring through digital mediums,
which is an increasing phenomenon as firms implement
technology in the service process and recognize social
media as part of the broadened servicescape. The extent
a digital medium facilitates both the affective and infer-
ential pathways of the EASI model should determine the
likelihood of technology facilitating the positive influence
of employee satisfaction levels and trajectories. Research
indicates that customers infer employee attributes through
chats and online communications (Li et al. 2019)butthe
ability to perceive changes in employee attributes is un-
known. Stated another way, can customers infer improv-
ing employee satisfaction through digital chat conversa-
tions? If not, then the increasing infusion of technology
into the service encounter may hinder the benefit of im-
proving employee happiness. Thus, future research should
test the current model across different medium contexts to
find important technological boundary conditions.
J. of the Acad. Mark. Sci.
Appendix 1
Appendix 2
Table 5 Correlation matrix for Study 1
MSD12345678910111213
1 Employee satisfaction
t0
3.16 0.45
2 Employee satisfaction
t1
3.02 0.46 .64
3 Employee satisfaction
t2
2.91 0.46 .43 .52
4 Employee satisfaction
t3
2.88 0.45 .49 .50 .55
5 Customer satisfaction
t3
82.49 10.80 .13 .16 .04 .10
6 Customer satisfaction
t4
75.06 5.32 .19 .25 .18 .19 .86
7 Contact 0.41 0.49 −.23 −.34 −.27 −.35 .05 −.02
8 ln (Company size) 4.65 1.79 −.10 −.07 −.00 .05 −.10 −.07 .23
9 ln (Company relative size) 2.03 1.26 .09 .17 .31 .32 −.05 −.09 −.19 .45
10 ln (Number of competitors) 2.33 0.65 .04 .07 −.08 −.07 .14 .17 .34 −.14 −.43
11 Downsizing (0 = no, 1 = yes) 0.09 0.28 −.06 −.12 −.11 −.12 −.10 −.11 .15 .19 .06 .06
12 Total prior downsizings 0.37 0.79 −.08 −.16 −.13 −.12 −.11 −.11 .11 .18 .10 −.02 .68
13 Financial loss (0 = no, 1 = yes) 0.03 0.17 −.10 −.10 −.06 −.07 −.05 −.07 .06 .06 .01 −.05 .16 .29
14 Total prior financial losses 0.12 0.55 −.09 −.12 −.10 −.08 −.07 −.09 .06 .05 −.00 −.09 .14 .23 .79
All correlations with an absolute value greater than .11 are significant at a pvalue of .05. The correlations for customer satisfaction
t3
are based on the
reduced sample size whereas all other correlations are based on the full sample size
Table 6 Survey measures for
Study 2 SL AVE αCR
Employee satisfaction (completely disagree / completely agree, 5-point scale) .74 .93 .93
1. I would recommend this Company to a friend as a good place to work.
2. Management is focused on the long-term success of my organization
3. My manager acts on my suggestions.
4. I have opportunities to receive training which helps me develop my skills.
5. Our strategies will make us more successful over the long term.
.87
.86
.83
.81
.91
Repatronage intentions (not at all likely/extremely likely; 5-point scale)
1. How likely are you to rent in the future?
Value perceptions (poor / excellent, 10-point scale)
2. Please rate your experience with us in the following areas: Value for the money
SL standardized loading, AVE average variance extracted, αCronbach’s alpha, CR composite reliability
J. of the Acad. Mark. Sci.
Appendix 3
References
Anderson, E. W., Fornell, C., & Lehmann, D. R. (1994). Customer satis-
faction, market share, and profitability: Findings from Sweden.
JournalofMarketing,58(3), 53–66.
Anderson, E. W., Fornell, C., & Rust, R. T. (1997). Customer satisfaction,
productivity, and profitability: Differences between goods and ser-
vices. Marketing Science, 16(2), 129–145.
Asparouhov, T., Hamaker, E. L., & Muthén, B. (2018). Dynamic struc-
tural equation models. Structural Equation Modeling: A
Multidisciplinary Journal, 25(3), 359–388.
Babin, B. J., Griffin, M., & Babin, L. (1994). The effect of motivation to
process on consumers' satisfaction reactions. In C. T. Allen & D. R.
John (Eds.), Na - advances in consumer research (pp. 406–411).
Provo: Association for Consumer Research.
Bandura, A. (1971). Social learning theory. New York: General Learning
Press.
Beer, M. (1964). Organizational size and job satisfaction. Academy of
Management Journal, 7(1), 34–44.
Bentein, K., Vandenberghe, C., Vandenberg, R., & Stinglhamber, F.
(2005). The role of change in the relationship between commitment
and turnover: A latent growth modeling approach. Journal of
Applied Psychology, 90(3), 468–482.
Bhattacharya, C. B., & Sen, S. (2003). Consumer–company identifica-
tion: A framework for understanding consumers' relationships with
companies. Journal of Marketing, 67(2), 76–88.
Bitner, M. J. (1995). Building service relationships: It’s all about prom-
ises. Journal of the Academy of Marketing Science, 23(4), 246–251.
Bollen, K. A., & Curran, P. J. (2006). Latent curve models: A structural
equation perspective. Hoboken: John Wiley & Sons, Inc.
Bosco, F. A., Aguinis, H., Singh, K., Field, J. G., & Pierce, C. A. (2015).
Correlational effect size benchmarks. Journal of Applied
Psychology, 100(2), 431–449.
Brown, S. P., & Lam, S. K. (2008). A meta-analysis of relationships
linking employee satisfaction to customer responses. Journal of
Retailing, 84(3), 243–255.
Celsi, R. L., & Olson, J. C. (1988). The role of involvement in attention
and comprehension processes. Journal of Consumer Research,
15(2), 210–224.
Certo, S. T., Withers, M. C., & Semadeni, M. (2016). A tale of two
effects: Using longitudinal data to compare within- and between-
firm effects. Strategic Management Journal, 38(7), 1536–1556.
Chen, G., Ployhart, R., Thomas, H. C.,Anderson, N., & Bliese, P. (2011).
A power of momentum: A new model of dynamic relationships
between job satisfaction and turnover intentions. Academy of
Management Journal, 54(1), 159–181.
Croon, M. A., & Van Veldhoven, M. J. P. M. (2007). Predicting group-
level outcome variables from variables measured at the individual
level: A latent variable multilevel model. Psychological Methods,
12(1), 45–57.
Dietz, J., Pugh, S. D., & Wiley, J. W. (2004). Service climate effects on
customer attitudes: An examination of boundary conditions.
Academy of Management Journal, 47(1), 81–92.
Echambadi, R., Campbell, B., & Agarwal, R. (2006). Encouraging best
practice in quantitative management research: An incomplete list of
opportunities. Journal of Management Studies, 43(8), 1801–1820.
Evanschitzky, H., Wangenheim, F. V., & Wünderlich, N. V. (2012). Perils
of managing the service profit chain: The role of time lags and
feedback loops. Journal of Retailing, 88(3), 356–366.
Giebelhausen, M., Robinson, S. G., Sirianni, N. J., & Brady, M. K.
(2014). Touch vs. tech: When technology functions as a barrier or
Table 7 Correlation matrix for Study 2
M SD Level 2 correlations Level 1 correlations
12345678910111213
1 Employee satisfaction
t0
3.63 0.55
2 Employee satisfaction
t1
3.60 0.45 .64
3 Employee satisfaction
t2
3.91 0.47 .54 .66
4 Employee satisfaction
t3
3.99 0.49 .47 .63 .71
5 ln (Size of location) 5.18 0.71 −.19 −.20 −.32 −.25
6 Average employee tenure 2.76 0.35 −.13 −.12 −.26 −.12 .35
7 Growth of rental center 0.84 21.55 −.17 −.12 −.24 −.31 −.06 −.27
8 Future revenue 190.94 22.19 −.20 −.10 −.09 −.08 .19 .35 −.00
9 Store level repatronage
t3
3.61 0.27 −.01 .17 .13 .36 .17 .30 −.13 .18
10 Value perceptions 7.09 2.07 −.20 −.18 −.08 −.23 −.36 −.31 .38 −.38 .02
11 Repatronage intentions
t4
3.65 1.11 −.05 .09 .09 .32 .14 .32 −.14 .14 .96 .63
12 ln (Contact) 0.98 1.14 −.02 −.08 −.18 −.06 .10 .17 .07 .18 .11 .05 .09
13 ln (Rental cost) 5.01 1.18 −.42 −.39 −.29 −.36 .20 −.05 .12 .10 −.15 −.10 −.00 .01
14 Purpose (0 = business, 1 = leisure) 0.40 0.48 .13 −.02 −.07 .13 .11 −.02 .03 .14 .02 .05 .01 −.26 −.06
For Level 2 (Level 1), all correlations with an absolute value greater than .21 (.00) are significant at a pvalue of .05
J. of the Acad. Mark. Sci.
a benefit to service encounters. Journal of Marketing, 78(4), 113–
124.
Govind, R.,Chatterjee, R., & Mittal, V. (2018). Segmentation of spatially
dependent geographical units: Model and application. Management
Science, 64(4), 1941–1956.
Groth, M., Hennig-Thurau, T., & Walsh, G. (2009). Customer reactions to
emotional labor: The roles of employee acting strategies and cus-
tomer detection accuracy. Academy of Management Journal, 52(5),
958–974.
Habel, J., & Klarmann, M. (2015). Customer reactions to downsizing:
When and how is satisfaction affected? Journal of the Academy of
Marketing Science, 43(6), 768–789.
Harmeling, C. M., Palmatier, R. W., Houston, M. B., Arnold, M. J., &
Samaha, S. A. (2015). Transformational relationship events. Journal
of Marketing, 79(5), 39–62.
Harter, J. K., Schmidt, F. L., & Hayes, T. L. (2002). Business-unit-level
relationship between employee satisfaction, employee engagement,
and business outcomes: A meta-analysis. JournalofApplied
Psychology, 87(2), 268–279.
Heskett, J., Sasser, W. E., & Schlesinger, L. A. (1997). The service profit
chain: How leading companies link profit and growth to loyalty,
satisfaction and value. New York: Free Press.
Hobfoll, S. E. (1989). Conservation of resources: A new attempt at con-
ceptualizing stress. American Psychologist, 44(3), 513–524.
Hogreve, J., Iseke, A., Derfuss, K., & Eller, T. (2017). The service–profit
chain: A meta-analytic test of a comprehensive theoretical frame-
work. JournalofMarketing,81(3), 41–61.
Hollmann,T., Jarvis, C. B., & Bitner, M. J. (2015). Reaching the breaking
point: A dynamic process theory of business-to-business customer
defection. Journal of the Academy of Marketing Science, 43(2),
257–278.
Homburg, C., & Stock, R. (2004). The link between salespeople’sjob
satisfaction and customer satisfaction in a business-to-business con-
text: A dyadic analysis. Journal of the Academy of Marketing
Science, 32(2), 144–158.
Hsee, C. K., & Abelson, R. P. (1991). Velocity relation: Satisfaction as a
function of the first derivative of outcome over time. Journal of
Personality and Social Psychology, 60(3), 341–347.
Huang, M.-H., & Cheng, Z.-H. (2016). A longitudinal comparison of
customer satisfaction and customer-company identification in a ser-
vice context. Journal of Service Management, 27(5), 730–750.
Huang,M.,Li,P.,Meschke,F.,&Guthrie,J.P.(2015).Familyfirms,
employee satisfaction, and corporate performance. Journal of
Corporate Finance, 35,108–127.
Johnson, J., Tellis, G. J., & Macinnis, D. J. (2005). Losers, winners, and
biased trades. Journal of Consumer Research, 32(2), 324–329.
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of
decision under risk. Econometrica, 47(2), 263–291.
Keiningham, T. L., Aksoy, L., Cooil, B., Peterson, K., & Vavra, T. G.
(2006). A longitudinal examination of the asymmetric impact of
employee and customer satisfaction on retail sales. Managing
Service Quality: An International Journal, 16(5), 442–459.
Kelava, A., Moosbrugger, H., Dimitruk, P., & Schermelleh-Engel, K.
(2008). Multicollinearity and missing constraints: A comparison of
three approaches for the analysis of latent nonlinear effects.
Methodology European Journal of Research Methods for the
Behavioral and Social Sciences, 4(2), 51–66.
Killingsworth, M. A., & Gilbert, D. T. (2010). A wandering mind is an
unhappy mind. Science, 330(6006), 932–932.
Kinicki, A. J., Mckee-Ryan, F. M., Schriesheim, C. A., & Carson, K. P.
(2002). Assessing the construct validity of the job descriptive index:
A review and meta-analysis. Journal of Applied Psychology, 87(1),
14–32.
Kowske, B. (2012). Employee engagement: Market review, buyer’sguide
and provider profiles. http://www.bersin.com/engagement-market-
review. Accessed January 10, 2017.
Lebreton, J. M., & Senter, J. L. (2008). Answers to 20 questions about
interrater reliability and interrater agreement. Organizational
Research Methods, 11(4), 815–852.
Li, J., Burch, T. C., & Lee, T. W. (2017). Intra-individual variability in job
complexity over time: Examining the effect of job complexity tra-
jectory on employee job strain. Journal of Organizational Behavior,
38(5), 671–691.
Li, X. S., Chan, K. W., & Kim, S. (2019). Service with emoticons: How
customers interpret employee use of emoticons in online service
encounters. Journal of Consumer Research, 45(5), 973–987.
Liu, D., Mitchell, T., Lee, T., Holtom, B., & Hinkin, T. (2012). Employees
are out of step with coworkers: Job satisfaction trajectory and dis-
persion influence individual- and unit-level voluntary turnover.
Academy of Management Journal, 55(6), 1130–1380.
Maas, C. J. M., & Hox, J. J. (2005). Sufficient sample sizes for multilevel
modeling. Methodology European Journal of Research Methods for
the Behavioral and Social Sciences, 1(3), 86–92.
Mani, V., Kesavan, S., & Swaminathan, J. M. (2014). Estimating the
impact of understaffing on sales and profitability in retail stores.
Production and Operations Management, 24(2), 201–218.
Marinova, D., Singh, S. K., & Singh, J. (2018). Frontline problem-
solving effectiveness: A dynamic analysis of verbal and nonverbal
cues. Journal of Marketing Research, 55(2), 178–192.
Mayer, D. M., Ehrhart, M. G., & Schneider, B. (2009). Service attribute
boundary conditions of the service-climate satisfaction link.
Academy of Management Journal, 52(5), 1034–1050.
Melián-González, S., Bulchand-Gidumal, J., & López-Valcárcel, B. G.
(2015). New evidence of the relationship between employee satis-
faction and firm economic performance. Personnel Review, 44(6),
906–929.
Mende, M., & Van Doorn, J. (2014). Coproduction of transformative
services as a pathway to improved consumer well-being: Findings
from a longitudinal study on financial counseling. Journal of Service
Research, 18(3), 351–368.
Meyers-Levy, J., & Sternthal, B. (1993). A two-factor explanation of
assimilation and contrast effects. Journal of Marketing Research,
30(3), 359–368.
Morgan, N. A., & Rego, L. L. (2006). The value of different customer
satisfaction and loyalty metrics in predicting business performance.
Marketing Science, 25(5), 426–439.
Newman, D. A. (2003). Longitudinal modeling with randomly and sys-
tematically missing data: A simulation of ad hoc, maximum likeli-
hood, and multiple imputation techniques. Organizational Research
Methods, 6(3), 328–362.
Palmatier, R. W., Houston, M. B., Dant, R. P., & Grewal, D. (2013).
Relationship velocity: Toward a theory of relationship dynamics.
JournalofMarketing,77(1), 13–30.
Peters, K., & Kashima, Y. (2015). A multimodal theory of affect diffu-
sion. Psychological Bulletin, 141(5), 966–992.
Petty, R. E., Cacioppo, J. T., & Schumann, D. (1983). Centraland periph-
eral routes to advertising effectiveness: The moderating role of in-
volvement. Journal of Consumer Research, 10(2), 135–146.
Pugh, S. D. (2001). Service with a smile: Emotional contagion in the
service encounter. Academy of Management Journal, 44(5), 1018–
1027.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models.
Thousand Oaks: Sage Publications, Inc.
Ryan, A. M., Schmit, M. J., & Johnson, R. (1996). Attitudes and effec-
tiveness: Examining relations at an organizational level. Personnel
Psychology, 49(4), 853–882.
Scharitzer, D., & Korunka, C. (2000). New public management:
Evaluating the success of total quality management and change
management interventions in public services from the employees'
and customers' perspectives. Total Quality Management, 11(7),
941–953.
J. of the Acad. Mark. Sci.
Segrin, C. (2004). Concordance on negative emotion in close relation-
ships: Transmission of emotion or assortative mating? Journal of
Social and Clinical Psychology, 23(6), 836–856.
Sirdeshmukh, D., Singh, J., & Sabol, B. (2002). Consumer trust, value,
and loyalty in relational exchanges. Journal of Marketing, 66(1),
15–37.
Steenkamp, J.-B. E. M., & Baumgartner, H. (2000). On the use of struc-
tural equation models for marketing modeling. International
JournalofResearchinMarketing,17(2–3), 195–202.
Van Kleef, G. A. (2009). How emotions regulate social life: The emotions
as social information (easi) model. Current Directions in
Psychological Science, 18(3), 184–188.
Wangenheim, F. V., Evanschitzky, H., & Wunderlich, M. (2007). Does
the employee–customer satisfaction link hold for all employee
groups? Journal of Business Research, 60(7), 690–697.
Weber, L. (2016). Job satisfaction hits a 10-year high—But it’s still below
50%. http://www.wsj.com/articles/job-satisfaction-hits-a-10-year-
highbut-its-still-below-50-1468940401. Accessed August 23, 2016.
Ye, J., Marinova, D., & Singh, J. (2007). Strategic change implementation
and performance loss in the front lines. Journal of Marketing, 71(4),
156–171.
Ye, S., Ng, T. K., & Lam, C. L. (2018). Nostalgia and temporal life
satisfaction. Journal of Happiness Studies, 19(6), 1749–1762.
Zablah, A. R., Carlson, B. D., Donavan, D. T., Maxham, J. G., III, &
Brown, T. J. (2016). A cross-lagged test of the association between
customer satisfaction and employee job satisfaction in a relational
context. Journal of Applied Psychology, 101(5), 743–755.
Zhao, X., Lynch, J. J. G., & Chen, Q. (2010). Reconsidering baron and
Kenny: Myths and truths about mediation analysis. Journal of
Consumer Research, 37(2), 197–206.
Zheng, Z., Pavlou, P. A., & Gu, B. (2014). Latent growth modeling for
information systems: Theoretical extensions and practical applica-
tions. Information Systems Research, 25(3), 547–568.
Publisher’s note Springer Nature remains neutral with regard to jurisdic-
tional claims in published maps and institutional affiliations.
J. of the Acad. Mark. Sci.
