Download - Uyung-The Mediating Role of Trust
The effects of soft and hard service attributeson loyalty: the mediating role of trust
Seigyoung Auh
Department of Management, Marketing Section, The University of Melbourne, Australia
AbstractPurpose – Drawing on social exchange and similarity attraction theories, the purpose of this paper is to explore the effect of soft attributes (e.g. socialand relational attributes) on hard attributes (e.g. core attribute performance) in a high contact service context, namely in the hair care serviceenvironment.Design/methodology/approach – This research extends the key mediating variable model (KMVM) of Morgan and Hunt by hypothesising that, whiletrust fully mediates the effect of soft attributes on loyalty, trust only partially mediates the effect of hard attributes on loyalty. Data were collected usingthe critical incident technique from 176 students.Findings – Supports the fully mediating role of trust for soft attributes and a partially mediating role of trust for hard attributes.Research limitations/implications – Future research should test the model in contexts that involve less employee-customer contact and interaction.Originality/value – The study underscores the practical importance of investing in non-core (soft attributes such as social and relational attributes that focuson social bonding) attributes in addition to core attributes (hard attributes such as performance excellence of key service delivered) in service marketing.
Keywords Social interaction, Customer loyalty, Trust
Paper type Research paper
An executive summary for managers and executive
readers can be found at the end of this article.
Introduction
Despite the relevance and significance of non-core (soft
attributes such as social and relational attributes) and core
attributes (hard attributes such as performance excellence of
key service delivered) in service marketing, little research has
been addressed to draw any definitive conclusions as to the
relationship between these two types of attributes (Driver and
Johnston, 2001). Moreover, little do we know how these two
different kinds of attributes differentially influence trust and
loyalty within a system of cause and effect framework. To this
end, the goal of this research is to fill the void in the services
marketing literature on the above two research questions. In
other words, does social bonding between employees and
customers affect customer’s perceptions of the core service
performance and what is the process by which these social
bonding attributes such as employee-customer interaction
quality and employee-customer characteristic similarity
influence customers’ loyalty towards the service provider? Is
the process by which these social and relational attributes
affect loyalty different from the way core service performance
affects loyalty? We hope to address these inquiries from the
present study.
In service marketing, service providers deliver not only hard
attributes but also soft attributes such as ambient atmosphere,
close relationships with customers through interaction quality,
and identification through creating and sharing greater
similarities between employees and customers. Service
managers expect and anticipate that these soft attributes,
let alone the hard attributes, will exert a positive effect in
retaining customers particularly in high contact and interaction
service contexts. For example, hair salons provide not only a
soothing and relaxing atmosphere to their customers but also
train their employees to identify with and share greater
similarities with their customers to continuously gain business.
The successful coffee franchise, Starbucks, continuously
innovates to deliver an outstanding coffee consumption
experience by providing a relaxing atmosphere whereby
customers can unwind and engage in a comfortable
conversation (Kachra and Crossan, 1997). They also go out
of the way to train their employees to obtain a confident level of
coffee knowledge to engage in a quality interaction and share
commonalities of the coffee consumption experience,
especially, with coffee savvy customers.Drawing from social exchange theory and similarity
attraction theory, this paper explores the effect of soft
attributes (e.g. social and relational attributes that focus on
social bonding) on hard attributes (e.g. core attribute
performance) in a high contact service context, namely in
the hair care service industry. This research also extends the
key mediating variable model (KMVM) of Morgan and Hunt
(1994) into the service sector by hypothesising that while core
attribute performance and trust fully mediate the effect of soft
attributes on loyalty, trust only partially mediates the effect of
hard attributes on loyalty suggesting that hard attributes can
directly impact loyalty.Morgan and Hunt (1994) in their seminal paper show that
trust and commitment are key mediating variables (KMV) in
a business-to-business relationship marketing context. We
extend on this notion of trust as a key mediator and posit that
The Emerald Research Register for this journal is available at
www.emeraldinsight.com/researchregister
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/0887-6045.htm
Journal of Services Marketing
19/2 (2005) 81–92
q Emerald Group Publishing Limited [ISSN 0887-6045]
[DOI 10.1108/08876040510591394]
81
the extent to which it functions as a mediator may depend on
the type of antecedent under consideration. Although
accumulated research exists especially in the channel andbusiness-to-business context (Anderson and Weitz, 1989;
Doney and Cannon, 1997; Ganesan, 1994; Kumar et al.,1995; Morgan and Hunt, 1994), relatively scant findings areavailable in the service context (see however Crosby et al.,1990; Moorman et al., 1992, 1993 for exceptions) on the
different types of antecedents of trust and how these may havea direct or indirect effect on loyalty via trust.The result of this research is expected to contribute to our
understanding of the relationship between soft and hard
attributes. As a result, managers can better allocate resources
effectively and appropriately by knowing that the performanceof soft attributes positively influences the perceived
performance of hard attributes. On the other hand, this
paper is also expected to shed light on the limitation of softattributes versus hard attributes in influencing loyalty via the
extent to which trust is a mediator between the two types of
attributes and loyalty.This paper is organised as follows. The next section
discusses the theoretical background, social exchange theoryand similarity attraction theory, and the research hypotheses.
This is followed by the empirical results of the conceptual
model used to test the hypotheses. Finally, theoreticalcontributions, managerial implications, limitations, and
directions for future studies are discussed.
Theoretical background and hypotheses
We argue in this paper that there are two types of attributes that
contribute to developing trust (Driver and Johnston, 2001).
Driver and Johnston (2001) report that predictors of servicequality can be broadly categorised into interpersonal (soft) and
non-interpersonal (hard) attributes. They extend the work of
Johnston (1995) who shows an exhaustive list of service qualitydrivers. Some of the attributes classified as soft attributes are
those that involve more personal or human contact and
interaction such as attentiveness/helpfulness, care, courtesy, andfriendliness. On the other hand, hard attributes comprise those
such as competence, functionality, and reliability. Driver and
Johnston’s (2001) work provide support for relaters placingmore importance on soft attributes and non-relaters putting
greater emphasis on hard attributes in determining their service
quality decisions. We next discuss the details of the soft andhard attributes used in our study.Within soft attributes, a distinction is made between
interaction quality that underscores the relational and social
connectedness between employees and customers and
characteristic similarity that contribute in social bondingwith customers (i.e. human); and atmosphere of the store (i.e.
non-human). Conversely, hard attributes are mainly
concerned with the quality of the core service providedwhich we refer to henceforth as perceived performance
excellence (PPE). According to our conceptual model
depicted in Figure 1, we propose two types of humanrelated soft attributes, interaction quality and characteristic
similarity with the service provider, and one non-human
related soft attribute, store atmosphere.Drawing on the principle of reciprocity, social exchange
theory, and the similarity attraction paradigm, these softattributes are expected to positively affect PPE that we use as
the hard attribute in our model. These soft attributes are also
expected to have a positive effect on trust and an indirect effect
on loyalty via PPE and trust. In other words, the soft attributesdo not have a direct effect on loyalty but only an indirect effect
through PPE and trust. On the other hand, PPE is predicted tohave a direct effect on loyalty in addition to an indirect effect on
loyalty though trust. Loyalty in this research is consistent with
behavioural intention and captures a customer’s predispositionto repurchase the service offering (Boulding et al., 1993; Dick
and Basu, 1994; Parasuraman et al., 1991). Our loyaltyconstruct also takes into account the resilience of this
predisposition in the face of price increase (Fornell et al., 1996).The distinction between soft and hard attributes is
consistent with what Gronroos (1984) asserts as technicalquality and functional quality. Similar arguments have been
proposed in the service marketing literature for a distinctionbetween functional or core service and relational or social
aspects of a service (Goodwin and Gremler, 1996; Hennig-Thurau et al., 2000). In the following sections, we discuss the
relationships between the constructs of our conceptual modeland derive relevant hypotheses.
Interaction quality
Social exchange theory argues that resources, tangible and
intangible, are exchanged between individuals or betweengroups with the goal of enhancing, maintaining, or
dismantling relationships or interactions. This theory hasbeen used extensively to explain intimate interpersonal
relationships in the family and social psychology literature(Burgess and Huston, 1979; Floyd and Wasner, 1994). In a
business context, social exchange theory has been adopted toexplain word of mouth communication (Frenzen and
Nakamoto, 1993; Gatignon and Robertson, 1986). Theconcept of relationship closeness was investigated using social
exchange theory in an industrial buyer-seller relationship. Theidea of exchange theory is not totally new to marketing in that
Bagozzi (1975) defined marketing as exchange.In this paper, social exchange theory is used to explain the
relationship between interaction quality and PPE andbetween interaction quality and trust. Interaction quality is
defined as the positive outcome derived from exchange ofconversational and verbal information between the hair care
provider and the customer. More often than not, during ahaircut experience, the customer engages in some kind of
interaction with the hair care provider. The quality and depthof this conversational interaction will obviously range from
mere shallow greetings to in depth personal conversations.Thus, this construct captures the quality of interaction
engagement experienced by the customer from a dialogue
with the hair care provider. Albeit similar to the well knownrelationship quality construct that encompasses the
relationship derived from various and broad perspectives,this interaction quality construct is narrower in scope and
particularly taps into the quality of the relationship built fromengaging in a conversational interaction.Social exchange theory suggests that reciprocal exchange of
positive and valuable information enhances the trust,
commitment and satisfaction with the relationship. Frequentand healthy exchange is expected to bring the parties closer to
one another and foster a lasting relationship. Moreover, greaterinteraction quality is expected to facilitate interpersonal and
social bonding, which in turn is also expected to foster trust.This implies that greater interaction quality will breed greater
trust. Moreover, due to the closeness and elevated relationship
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
82
that is expected to be cultivated from greater interaction quality,
this is posited to enhance the perception of the core service
performance or PPE. In the context of this study, PPE is
defined as the customer’s perception of the hair care provider’s
ability to perform a quality haircut. This indicates the core
service that every hair salon tries to satisfactorily deliver. In the
conceptual model, this PPE refers to the hard attribute or core
service. Greater interaction quality is expected to increase the
social and interpersonal bonding between the customer and the
hair care provider, resulting in greater trust and PPE.The economic sociology literature suggests that within an
exchange relationship, a party derives utility not only from the
products and services exchanged but also from interpersonal
relationships (Wathne et al., 2001; Frenzen and Davis, 1990).
Thus, to the extent that a hair care provider makes efforts to
show care and benevolence towards their customers, this is
expected to foster the creation of psychological bonds that are
foundations for strong relationships (De Wulf et al., 2001;
Crosby et al., 1990). When customers perceive that their hair
care providers are exerting efforts to enhance interaction
quality, they often reciprocate in kind with good will or in our
case by showing greater PPE. Bagozzi (1995) argued that
according to the principle of reciprocity, people should return
good for good in proportion to what they receive. Therefore
the greater the customer feels that interaction quality is being
met, the more favourably the customer is expected to perceive
the performance of the core service delivery.Indirect support for a positive effect from interaction
quality to PPE and trust comes from the work of Gwinner
et al. (1998) who identified three types of relational benefits
(e.g. confidence, social, and special treatment benefits) that
contribute in maintaining a strong relationship between
customers and service providers. Furthermore, Beatty et al.
(1996) and Reynolds and Beatty (1999) identified two
primary types of relational benefits derived from an exchange
relationship, functional benefits and social benefits. Based on
these categorisations, it has been shown that social benefits
have a positive effect on functional benefits and satisfaction
(Gremler and Gwinner, 2000; Hennig-Thurau et al., 2002;
Price and Arnould, 1999; Reynolds and Beatty, 1999). It has
also been shown that rapport (in particular, enjoyable
interaction) positively influenced satisfaction with the service
(Gremler and Gwinner, 2000). The work of Goodman et al.
(1995) is also consistent with our argument. They found that
the performance of peripheral factors had an effect on the
perceived performance of core factors. Their results suggested
that the performance of a peripheral factor (satisfaction with
responsiveness of inquiry handling) not only had a direct
effect on overall customer satisfaction but also an indirect
effect through its effect on perceived performance of a core
factor (satisfaction with first class mail).Although interaction quality is expected to have a direct
effect on PPE and trust, it is not expected to directly affect
loyalty. We expect interaction quality to influence loyalty
indirectly through PPE and trust. Generally speaking, the
likelihood that customers will visit their hair care providers
simply based on satisfactory interaction quality is slim. The
main impetus for patronage would still be based on the
performance of the haircut and the trust that customers have
towards their hair care providers in expectation of receiving a
reliable and benevolent service during their next visit. The
above arguments lead to the following hypotheses about
interaction quality.
H1. Greater interaction quality is associated with greater
PPE.
H1b. Greater interaction quality is associated with greater
trust.
H1c. Interaction quality does not have a direct effect on
loyalty but only an indirect effect through PPE and trust.
Characteristic similarity
Research in industrial and organisational psychology has usedthe similarity attraction paradigm to explain the effect of
homogeneous groups on work performance and turnover(O’Reilly et al., 1989). Social categorisation theory (Tajfel andTurner, 1986) suggests that people with greater similarity in
social categories such as age, sex, race, and status interactmore frequently with in-group than with out-group members.
Figure 1 Proposed conceptual model
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
83
These arguments are consistent with the well known adage,
“birds of a feather flock together”. Therefore it can be
inferred from the similarity-attraction paradigm (Baskett,
1973; Byrne, 1971) that people have the inclination to prefer
and feel more comfortable with people of similar personalities
and characteristics. In this paper, the construct of
characteristic similarity is employed to convey the notion of
similarity that the customer has with the hair care provider in
terms of tastes, conversation topic, and in more general,
simply similar people. Put differently, characteristic similarity
is about the commonality that the customer shares with its
hair care provider.In similar light, greater homophily (Ibarra, 1992) and similar
relational demography (Tsui and O’Reilly, 1989) is reasoned to
contribute to greater attraction and preference. Nicholson et al.(2001) provided empirical support for the similarity-attraction
theory. Their research showed that similarity of business value
between a buyer and a sales representative had a positive effect
on the buyer’s liking for the sales representative. In a buyer-
seller context, Smith (1998) reported that similarity in life-
stage, sex, cultural background, work attitude, and personality
had an indirect positive effect on relationship quality through
relational management.Along similar lines, Boles et al. (2000) and Crosby et al.
(1990) commonly found that the similarity between customer
and salesperson did not have a positive effect on relationship
quality. A possible nonsignificant result may be due to the fact
that people have an inherently low level of trust and
commitment towards sales people in general and required
more than just similarity to conclude that a quality
relationship exists. However, more recently, characteristics
of the salesperson relationship such as likeability and
similarity have proven to be significant antecedents to the
buying firm’s trust in the salesperson (Doney and Cannon,
1997). Thus, mixed results exist in the literature on the effect
of similarity on trust. However, both the Boles et al. (2000)
and Crosby et al. (1990) studies revealed that as the customer
perceived greater similarity between the two parties, the
salesperson was perceived to be more effective.Similar to the argument advanced for interaction quality’s
indirect effect on loyalty, we also posit that characteristic
similarity will only impact loyalty indirectly through PPE and
trust. It seems very unlikely that just because customers
perceive their hair care providers to be similar to them in
terms of taste and conversational topics, they will continue to
visit them. The dominant factor that drives customers’ loyalty
towards their hair care providers should be foremost the
quality of the haircut and the trust towards them. These two
elements should take precedence over characteristic similarity.
Therefore, greater characteristic similarity can have a direct
positive effect on PPE and trust but only an indirect effect on
loyalty via these two mediating constructs.Taken collectively, the above arguments contend that
greater similarity between the customer and the hair care
provider should enhance attraction and communication
resulting in a higher level of trust and PPE. Relationships
based on similar values and interests are likely to foster greater
identification with one another and build the platform for
trust. Based on the above arguments, we propose the
following hypotheses about characteristic similarity.
H2. Greater characteristics similarity is associated with
greater PPE.
H2b. Greater characteristics similarity is associated withgreater trust.
H2c. Characteristics similarity does not have a direct effecton loyalty but only an indirect effect through PPE andtrust.
Store atmosphere
The effect of the physical atmosphere and environment onhuman behaviour has been scrutinised in the environmentalpsychology literature (Mehrabian and Russell, 1974). Inmarketing, this concept has been successfully applied in retailsettings to show that pleasure, arousal, scent, and music canall have main and interactive effects in influencing consumerbehaviour in retail outlets (Donavan and Rossiter, 1982;Donavan et al., 1994; Dube et al., 1995, Wakefield and Baker,1998; Mattila and Wirtz, 2001). The work of Mattila andWirtz (2001) suggested that when congruency as opposed toincongruency existed between type of music and arousal levelof scent, customer’s shopping experience was enhanced.In addition to our human related soft attributes (i.e.
interaction quality and characteristic similarity), we had ameasure of the store atmosphere that we argued previously asa non-human related soft attribute. In other words, althoughnot part of the core service delivery, store atmosphere, alongwith interaction quality and characteristic similarity, isexpected to have a positive effect on PPE. More specifically,the expectation is that as customers more readily identify withthe store atmosphere, the more positive they are to perceivethe performance of the core service delivery (e.g. haircut).The rationale for this follows from the fact that the morecustomers can relate to and identify with their surroundingsand climate, the more comfortable they will feel and hencethis positive feeling is expected to carry over into theevaluations of the core service performance. Based on theabove argument, the following is proposed.
H3. Greater identification with the store atmosphere isassociated with greater PPE.
PPE and trust
Although trust has been defined in several different ways, weemploy the definition of Doney and Cannon (1997) who havedefined trust as the perceived credibility (reliability) andbenevolence (genuinely interested in other parties interest)toward the target of interest. Although this definition assumestrust as a multi-dimensional construct, empirically trust hasproven to be a global unidimensional construct (Doney andCannon, 1997; Joshi and Stump, 1999; Nicholson et al., 2001).The unidimensionality of trust is consistent with others whoview the credibility and benevolence dimensions as antecedentsrather than components of trust (Mayer et al., 1995).Consistent with the view that credibility (reliability) towards
the service provider is an antecedent rather than a componentof trust (Mayer et al., 1995) and drawing on the consequencesof service quality (Brady and Cronin, 2001), it is hypothesisedthat PPE or the perceived ability of the hair care provider todeliver a reliable and quality haircut is expected to enhancethe level of trust and loyalty towards the hair care provider.We also posit a direct effect of PPE on loyalty in addition toan indirect effect via trust. Consistent with our argument,Gruen et al. (2000) in a professional association context foundthat core service performance had a direct and positive effect
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
84
on retention. Based on the above reasoning, the followinghypotheses are proposed.
H4. Greater PPE is associated with greater trust.H5. Greater PPE is associated with greater loyalty not only
indirectly through trust but also directly.
Trust is believed to be a critical variable that contributes tohealthy and long-term relationships (Anderson and Weitz,1989; Doney and Cannon, 1997; Garbarino and Johnson,1999; Ganesan, 1994; Morgan and Hunt, 1994). We expectthis relationship to be even stronger in service contexts thatare characterised by high and close employee-customercontact such as that in the hair care service industry(Bowen, 1990). Bowen (1990) describes a taxonomy ofservice types whereby hair care service falls into the groupcharacterised as high interaction and contact. Furthermore,we expect trust to be positively linked to loyalty especiallywhen the service process is not standardised and a certainlevel of uncertainty is present that adds to the risk of beingsusceptible to vulnerability (e.g. unexpected and bad haircutthat the customer will have to live with until the next cut)(Deutsch, 1962; Moorman et al., 1992). Thus, we expecttrust to be positively associated with loyalty.
H6. Trust has a positive effect on loyalty.
Method
The hypotheses were tested using a survey with hair careservices. Hair care service was selected for the followingreasons. First, the nature of the service category carries acertain level of employee-customer interaction in terms ofsocial bonding that is key to testing our conceptual model.Second, it has been used in past research on relationshipbuilding (Gwinner et al., 1998; Hennig-Thurau et al., 2002;Price and Arnould, 1999; Jones et al., 2002). Third, hair careservice was a category that was sufficiently familiar for oursurvey respondents (college students) to provide reliable andquality data. Finally, it was a service category that both malesand females used on a relatively consistent basis such that ourempirical results would not be biased toward either gender.The attributes and benefit categories used in the survey
were obtained using CIT, the critical incident technique(Bitner et al., 1990; Grove and Fisk, 1997; Hayes, 1998;Meuter et al., 2000). The CIT approach focuses on inputfrom the customer to develop survey items (for a detailedexplanation of the CIT, refer to Hayes, 1998). A conveniencesample of ten doctoral students was asked to list five to tenpositive experiences (likes) as well as five to ten negativeexperiences (dislikes) regarding their past or current hair careservice provider. The interviews lasted approximately 15-20minutes and were audio-taped for subsequent analysis. A totalof 100 different critical incidents were obtained from these teninterviewees, ranging from 8 to 24 per interview. The criticalincidents were then categorised by the author and a colleagueinto 39 distinct performance attributes which, in turn, werefurther categorised into the benefit-level constructs. Theclassifications were highly reliable (over 80 per cent agreementfor both attributes and benefits) and disagreements wereresolved through discussion. The benefits served as theantecedents of trust in our conceptual model. It should benoted that the attributes were identified by customersthemselves rather than by the researchers.
Table I shows the 13 attribute performance measures used to
operationalise the benefits as latent variable antecedents to trustin our model. The benefit items were constructed using ten-
point disagree to agree scales. Trust and loyalty were also
operationalised as latent variables using the measures listed inTable I. They were collected using ten-point semantic
differential and likelihood scales. The self-administered surveywas completed by 191 undergraduate business students who
earned course credit for participating. A total of 12 subjectswere deleted from the analysis due to an excessive number of
missing values. An additional three subjects were removedbecause they indicated that a family member or friend acted as
their hair care provider. Consequently, 176 subjects wereincluded in the final analysis. Males and females were about
equally represented in our sample, with 51 per cent male and49 per cent female subjects.
Results
Measurement model
Lisrel 8.50 was used for all of the analysis unless specified
otherwise. We first assessed the measurement model followed
by the structural model for hypotheses testing following theguidelines suggested by Anderson and Gerbing (1988). We first
report the results of our confirmatory factor analysis (CFA).According to our CFA results, our measurement model
suggested good fit to the data (x2ð120Þ ¼ 290:68, p , 0:001,CFI ¼ 0:93, NNFI ¼ 0:91, RMSEA ¼ 0:082). The results of
the CFA with factor loadings and t-values are summarised inTable I. All factor loadings were high and significant,
providing strong evidence for convergent validity (Bagozziand Yi, 1988). We also assessed coefficient alpha (Cronbach,
1951) and composite reliability for each of the constructs(Fornell and Larcker, 1981). Coefficient alpha ranged from
0.75 (loyalty) to 0.91 (PPE) whereas the composite reliabilityranged from 0.76 (loyalty) to 0.91 (PPE and interaction
quality). Table II summarises the results of the reliability tests.Discriminant validity was assessed in three different ways.
The first method involved constructing a 95 per cent
confidence interval around the correlation of any twoconstructs (Anderson and Gerbing, 1988). Discriminant
validity is demonstrated when this confidence interval doesnot include 1. According to this confidence interval test, none
of the confidence intervals between any two constructsincluded 1 indicating support for discriminant validity. Our
second test of discriminant validity concerned conductingmultiple chi-square difference tests (Bagozzi and Phillips,
1982). This test involves taking the chi-square differencebetween an unconstrained model (whereby the correlation
between constructs is freely estimated) and a constrainedmodel (whereby the correlation between constructs of interest
is fixed to unity). Discriminant validity is demonstrated when
the chi-square difference is significant implying that theconstrained model fits the data worse than the unconstrained
model. Table III summarises the results for all possible pairsof constructs. According to Table III, all of the chi-square
differences are significant, providing further evidence ofdiscriminant validity among the constructs.The last of the discriminant validity tests which is viewed as
the most conservative involves calculating the average
variance extracted (AVE) for any pair of constructs andcomparing this to the squared correlation between the two
constructs (Fornell and Larcker, 1981). All AVE exceeded the
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
85
recommended level of 0.50 and ranged from 0.61 (customer
loyalty) to 0.78 (interaction quality) (Fornell and Larcker,
1981). With the exception of the discriminant validity
between trust and loyalty, all AVEs were greater than the
squared correlation between the two constructs of interest
(refer to Table II for details). However, as mentioned before,
this test is the most conservative and since discriminant
validity was established based on the prior two tests, we
cautiously proceeded with further analysis. Based on our
reliability, convergent, and discriminant validity tests, we
concluded that our measurement model met most of the
psychometric property requirements.
Structural model
We first report the fit of our proposed theoretical model to the
data. Several fit indices suggested good fit of our model to the
data (x2ð123Þ ¼ 290:38, p , 0:001, CFI ¼ 0:93, NNFI ¼ 0:91,RMSEA ¼ 0:08). One test that is very informative but often
overlooked is the comparison of fit between our proposed
theoretical model to that of our measurement model. A
Table I Items and measurement model results
Construct Description Loadings T-value�
Interaction quality 1 (strongly disagree) to 10 (strongly agree)a
IQ1 My hair care provider remembers my name 0.95 16.64
IQ2 My hair care provider remembers things about me 0.97 17.19
IQ3 I enjoy the topics that my hair care provider and I talk about 0.70 10.54
Characteristic similarity 1 (strongly disagree) to 10 (strongly agree)
CS1 My hair care provider and I are similar people 0.87 13.82
CS2 My hair care provider and I have similar tastes 0.91 14.79
CS3 My hair care provider and I like to talk about the same topics 0.78 11.88
Perceived performance excellence 1 (strongly disagree) to 10 (strongly agree)
PPE1 My hair care provider cuts my hair evenly 0.72 10.74
PPE2 My hair care provider is precise 0.90 15.10
PPE3 My hair care provider is skilled 0.90 15.03
PPE4 My hair care provider is careful 0.88 14.63
Store atmosphere 1 (strongly disagree) to 10 (strongly agree)
SA1 I have fun while I am at the shop 0.83 12.65
SA2 I like the general atmosphere in the shop 0.94 15.06
SA3 I can identify with the atmosphere in my hair care provider’s shop 0.65 9.29
Trust 1 (never) to 10 (always)
T1 The extent to which you are willing to rely on your hair care provider 0.83 13.08
T2 The extent to which you trust your hair care provider 0.92 15.34
T3 The extent to which you are willing to follow your hair care provider’s
recommendations for new hair styles
0.62 8.73
Loyalty 1 (never) to 10 (always)
L1 The likelihood that you will visit your hair care provider again 0.70 9.96
L2 In the event that your hair care provider increased the price by 10 per cent, the
likelihood that you will visit your hair care provider again at the higher price
0.86 12.68
Notes: aA DK (don’t know) response was included in all the scales to accommodate situations for no experience; �All t-tests significant at p , 0:001
Table II Inter-correlation and scale analysis for constructs
Variable Coefficient a Composite reliabilitya AVEb IQ CS SA PPE TT LOY
Interaction quality (IQ) 0.90 0.91 0.78 1.00 0.12c 0.21 0.34 0.44 0.30
Characteristic similarity (CS) 0.88 0.89 0.73 0.35 1.00 0.23 0.24 0.25 0.18
Store atmosphere (SA) 0.83 0.85 0.67 0.46 0.48 1.00 0.22 0.24 0.20
Perceived performance excellence (PPE) 0.91 0.91 0.73 0.58 0.49 0.47 1.00 0.59 0.58
Trust (TT) 0.81 0.84 0.64 0.66 0.50 0.49 0.77 1.00 0.74
Loyalty (LOY) 0.75 0.76 0.61 0.55 0.43 0.45 0.76 0.86 1.00
Notes: All correlations are significant at p , 0:001 (two-tailed tests); aComposite reliability = ðSlxiÞ2=ðSlxiÞ
2 þ S Varð1iÞ;bAVE=Slx2
i =Slx2i þ SVarð1iÞ;
cShared variances are reported in the upper half of the matrix
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
86
finding of no significant difference between the two models
provides support for the nomological validity of our
theoretical model (Anderson and Gerbing, 1988). If the
theoretical model is successful in accounting for the observed
relations between the constructs, there will not be a significant
difference between the chi-square for the theoretical and
measurement models. This is because the measurement
model is in fact a saturated model in which all possible
specifications between constructs have been exhausted. In
contrast, our theoretical model is a constrained version of this
saturated model in that some of the relationships between
constructs have been fixed to zero. Thus, the intent of our
theoretical model was to come up with a model that was more
parsimonious than the measurement model, but at the same
time did nearly as good a job as accounting for the covariance
between the constructs. We see that this was indeed the case
because the fit of our theoretical model did not deteriorate
compared to our measurement model (measurement
model:x2ð120Þ ¼ 290:68, p , 0:001 vs theoretical model:
x2ð123Þ ¼ 290:38, p , 0:001, x2dð3Þ ¼ 0:30, p . 0:10).We now turn our attention to the results of our hypotheses
testing. Table IV summarises the results of our hypotheses
testing (H1a-H6). H1a-H1c tested the relationship between
interaction quality and PPE (H1a), interaction quality and trust
(H1b), and the indirect effect of interaction quality on loyalty
(H1c). The path coefficient between interaction quality and PPE
was positive and significant (H1a: g11 ¼ 0:40, p , 0:01).Therefore H1a was supported. The path from interaction
quality to trust was also positive and significant (H1b:g21 ¼ 0:30, p , 0:01). Finally, to test for the indirect effect of
interaction quality to loyalty, we ran a model that included the
additional path from interaction quality to loyalty. If a direct
effect exists between the two constructs, we would observe the
fit of this model to improve compared to our baseline model. On
the other hand, if no direct effect is present and only an indirect
effect is at work, the new model would not improve significantlybeyond our current baseline model despite adding the additional
path. Our results confirmed that adding the additional path
from interaction quality to loyalty did not improve the fit of themodel above and beyond the current model (x2ð123Þ ¼ 290:38,p , 0:001 vs x2ð122Þ ¼ 289:71, x2dð1Þ ¼ 0:67, p . 0:30).Therefore, H1c was strongly supported.Next, we report the results of H2a-H2c. H2a posited that
characteristics similarity would be positively associated with PPE.
As expected, the path from characteristics similarity to PPE waspositive and significant (H2a: g13 ¼ 0:26, p , 0:01). The path
from characteristics similarity to trust was also positive andsignificant, lending support for H2b (H2b: g23 ¼ 0:14, p , 0:05).For H2c, we tested the indirect effect of characteristics similarityon loyalty as we did for interaction quality. The model with the
added path from characteristics similarity to loyalty did not
significantly improve the fit of the model beyond the baselinemodel (x2ð123Þ ¼ 290:38, p , 0:001 vs x2ð122Þ ¼ 290:22,x2dð1Þ ¼ 0:16, p . 0:50). Therefore, as was the case forinteraction quality, we found strong support for an indirect
effect of characteristic similarity on loyalty through PPE and trust.Our next hypothesis involved the relationship between store
atmosphere and PPE. The path coefficient between these twoconstructs was as expected, positive and significant (H3:g12 ¼ 0:16, p ¼ 0:05). Thus, in addition to our human-related soft attributes that fostered interpersonal and social
bonding effects, store atmosphere also had a positive effect onhow customers perceived the performance of core services.
Although not hypothesised, we tested to see if adding a pathfrom store atmosphere to trust would improve the fit of our
model. The results indicated that the model would not be
improved by adding this additional path (x2ð123Þ ¼ 290:38, p ,
0:001 vs x2ð122Þ ¼ 289:83, x2dð1Þ ¼ 0:55, p . 0:30) providing
support for our conceptual model.H4 and H5 are concerned with the consequences of PPE. In
particular, H4 stated that PPE would be positively associatedwith trust. Results confirmed our hypothesis in that the path
coefficient between PPE and trust was positive and significant(H4: b21 ¼ 0:53, p , 0:01). Furthermore, PPE was positively
associated with loyalty (H5: b31 ¼ 0:27, p , 0:05). H5 alsoargued that in addition to PPE having an indirect effect on
loyalty via trust, it would also have a direct effect. This wastested by running a model that deleted the path from PPE to
loyalty and comparing it with the baseline model. If a directeffect were to be supported we would be likely to observe a
significant deterioration in fit when the direct path from PPE to
loyalty was deleted. Results confirmed our expectations in thatwhen the direct path was deleted the fit became significantly
worse (x2ð123Þ ¼ 290:38, p , 0:001 vs x2ð124Þ ¼ 296:00,x2dð1Þ ¼ 5:62, p , 0:02). Therefore, unlike interaction quality
and characteristics similarity that had only an indirect effect onloyalty through PPE and trust, PPE had both an indirect effect
and also a direct effect on loyalty. Lastly, we report the effect oftrust on loyalty. The path coefficient between these two
constructs was positive and significant, providing support forH6 (H6: b32 ¼ 0:64, p , 0:01).
Conclusion
Theoretical contribution
The results of our study make two new contributions to the
relationship and service marketing literature. The first is the
Table III Discriminant validity test
Target x2(UM)b x2(CM)c Dx2d
F(j1, j2)a 290.68 310.57 19.89
F(j1, j3) 290.68 315.42 24.74
F(j1, j4) 290.68 338.67 47.99
F(j1, j5) 290.68 308.93 18.25
F(j1, j6) 290.68 324.01 33.33
F(j2, j3) 290.68 323.07 32.39
F(j2, j4) 290.68 350.43 59.75
F(j2, j5) 290.68 323.36 32.68
F(j2, j6) 290.68 336.40 45.72
F(j3, j4) 290.68 348.58 57.9
F(j3, j5) 290.68 321.41 30.73
F(j3, j6) 290.68 333.16 42.48
F(j4, j5) 290.68 334.74 44.06
F(j4, j6) 290.68 336.44 45.76
F(j5, j6) 290.68 329.74 39.06
Notes: aConstructs: trust (j1), loyalty (j2), perceived performanceexcellence (j3), store atmosphere (j4), interaction quality (j5),characteristic similarity (j6); bUnconstrained model whereby allcorrelations between constructs are freely estimated (df = 120);cConstrained model whereby the correlation between constructs incolumn 1 are fixed to unity in the phi matrix (df = 121); dSince theconstrained model is nested in the unconstrained model, the chi-squaredifference is 1 df and the critical value is Dx2 .10.83, p , 0.001
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
87
revelation of a positive link between soft attributes such asinteraction quality and characteristics similarity and hard
attributes such as PPE. The second is the degree to whichtrust mediates these soft and hard attributes on loyalty. Our
empirical results found that interaction quality and
characteristics similarity only had an indirect effect onloyalty that was mediated by PPE and trust. On the other
hand, PPE had both an indirect effect and also a direct effecton loyalty bypassing trust as a mediator. Subsequently, we
discuss each of the two contributions in greater detail.Although past research has shown different types of factors
that determine service quality (Driver and Johnston, 2001)
and the different types of benefits that customers can expectto receive from a relationship with service providers (Beatty
et al., 1996; Gwinner et al., 1998; Reynolds and Beatty,1999), little has been written to show the association between
interpersonal factors such as that between employees andcustomers and the performance of core services. Our results
are also in line with what Goodman et al. (1995) found. Their
findings provided support for the performance of a peripheral(i.e. non-core) service domain having a positive effect on the
performance of a core service domain. Results of our studyadd to the understanding of how different types of service
attributes are inter-related in a high-contact/interactionenvironment between employees and customers.Our positive effect from interaction quality and
characteristics similarity to PPE and trust can be interpreted
using the framework of social exchange theory and similarity
attraction paradigm. Social exchange theory argues thatreciprocal exchange of positive and valuable information
enhances the trust, commitment and satisfaction with therelationship. Positive exchange is expected to bring the parties
closer together and foster a bonding relationship. In otherwords, greater interaction quality is expected to bring about
interpersonal and social bonding, which in turn is also
expected to foster trust. The effect from interaction quality toPPE can be viewed in light of the principle of reciprocity
(Bagozzi, 1995). According to this principle, people have thetendency to return good for good in proportion to what they
receive. Therefore, the greater the customer feels that
interaction quality is fulfilled, the more favourably thecustomer is expected to perceive the performance of the
core service delivery.The results of the positive relationship between
characteristics similarity and PPE and trust are consistent
with the similarity attraction paradigm (Baskett, 1973; Byrne,1971) and with social categorisation theory (Tajfel and
Turner, 1986). These two theoretical arguments commonlypostulate that people have the tendency to gravitate towards
those of external and internal similarities. People like to be
around those of greater similarity because more often than notsimilarity breeds comfort, relatedness, and unity. As a
consequence, similarity contributes to trust and in-group vsout-group favourability.Our second contribution extends the results found in the
KMV model of Morgan and Hunt (1994) beyond a business-to-business marketing context to a service marketing arena.
The KMV model has shown that trust and commitment fullymediate the relationship between the antecedents and
consequences of trust and commitment. However, our model
supports the finding that while trust fully mediates the positiveeffect of interaction quality and characteristics similarity on
loyalty it only partially mediates the positive effect of PPE onloyalty. This adds significant richness to the KMV model by
specifying the conditions under which trust is a full or partial
mediator between its antecedents and consequences. This alsoextends our knowledge about the difference between soft and
human related attributes that bolster interpersonal and socialbonding compared to hard attributes that deal primarily with
the core service delivery per se.Our results are different from what Gwinner et al. (1998)
have found in this respect. In their study they found support
for social benefits to have a direct positive effect on loyalty.However, in their study they did not include trust in the
model. We contend that this may be one of the reasons for thisinconsistent result. When trust is added as an additional
construct into the model, personal and relational benefits
were found to have a positive effect on loyalty only throughtrust as our model suggested. Since these factors are personal
and relational in nature and have less to do with the actual
Table IV Structural model results
Hypothesis Coefficienta Chi-square difference test Fit indices
H1a: IQ!PPE (1 ) 0.40 (5.03) x2(123) ¼ 290.38
H1b: IQ ! TRUST (1 ) 0.30 (4.44) RMSEA ¼ 0.08
H1c: IQ ! LOYALTY (1 ) indirectly via PPE and TRUST x2(122)¼289.71b
x2d(1)¼0.67, p . 0.30
CFI ¼ 0.93
H2a: CS ! PPE (1 ) 0.26 (3.24) NNFI ¼ 0.91
H2b: CS ! TRUST (1 ) 0.14 (2.08)
H2c: CS ! LOYALTY (1 ) indirectly via PPE and TRUST x2(122)¼290.22c
x2d(1)¼0.16, p . 0.50
H3: SA ! PPE (1 ) 0.16 (1.96)
H4: PPE ! TRUST (1 ) 0.53 (6.05)
H5: PPE ! LOYALTY (1 ) 0.27 (2.42) x2(124)¼296.00d
x2d(1)¼5.62, p , 0.02
H6: TRUST ! LOYALTY (1 ) 0.64 (5.24)
Notes: IQ = interaction quality, PPE = perceived performance excellence, CS = characteristic similarity, SA = store atmosphere, RMSEA = root mean square errorof approximation, NNFI = non-normed fit index, CFI = comparative fit index; aValues in parentheses indicate t-values and all coefficients are significant at 0.05or less; bIndicates model where an additional path from IQ to loyalty was added to the baseline model; cIndicates model where an additional path from CS toloyalty was added to the baseline model; dIndicates model where the path from PPE to loyalty was deleted from the baseline model
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
88
performance of the core service activity per se, we would
expect the effect of these on loyalty to diminish significantly
once trust is included in the model. This finding is consistent
with the condition for a fully mediating effect suggested by
Baron and Kenney (1996).Taken collectively, our model has revealed that relational
and social factors as well as store atmosphere have direct
effects on PPE and trust but only an indirect effect on loyalty
through these mediating variables. Conversely, unlike the soft
attributes, core service performance factors such as PPE have
both an indirect effect and a direct effect on loyalty. Next we
look at the managerial relevance of our research.
Managerial implications
The results of our study suggest several implications to
managers who desire to build a loyal customer base based on
trust in a high-contact service context. The importance of
relational and social factors emerged as valuable avenues for
managers to allocate scarce resources. Having a service
environment that is conducive to similar characteristics and
social benefits can pay dividends toward increasing loyalty
through enhanced PPE and trust. Our findings are promising
for managers who desire, but at times are met with resistance
in the organisation, to invest in human resources to help
employees better interact and serve customers. Service
organisations, especially those that engage in a high level of
face-to-face contact and interaction with customers can
particularly benefit from the findings of our model.
Interaction quality and characteristics similarity are areas
where managers can consistently train and educate their
employees to “click” with their customers.Quite often managers tend to neglect the importance of
improving relationship building by focusing on only the
performance of core services. This may in part be due to
relational/social factors not having a direct effect on loyalty as
suggested by our model. These factors work their way towards
loyalty via improved PPE and trust. As a consequence,
managers overlook this opportunity and focus more
intensively on improving PPE as this, unlike relational/social
factors, has the tendency to have a direct effect on loyalty.
However, this may be overlooking and ignoring a potentially
fruitful and rewarding investment strategy that can have
promising payoffs. Whereas the performance of the core
service may be more difficult and costly to satisfactorily
deliver, relational and social benefits through mechanisms
such as interaction quality and characteristics similarity can
be implemented with far less resource and risk. Moreover,
under certain situations where it takes time to learn and
deliver a satisfactory core service, it may pay dividends for the
organisation to focus on the relational and social benefits to
retain customers while buying time to improve on its core
service performance.In this research, PPE was a unidimensional construct that
captured the customers’s perceived hair care provider’s ability
to deliver a quality hair cut. Our findings suggest that of the
three soft attributes, interaction quality had the greatest effect
on PPE followed by characteristic similarity and store
atmosphere. This provides managers to make effective
resource allocations to make strategic priority decisions on
the various soft attributes. Obviously, the high contact and
interactive nature of the hair care industry may have dictated
which soft attributes are more or less important in influencing
PPE and therefore the relative effect of these soft attributes on
PPE may change as a function of the service category.Finally, managers can also use relationship setup costs as a
means of increasing the perceived switching costs tocustomers (Jones et al., 2002). Customers may be able to
find a hair stylist that can do just as good a job if not better at
providing an excellent haircut but it may be more difficult tofind a hair stylist that they can connect with and build a
lasting relationship. Lastly, our positive effect from storeatmosphere to PPE replicates and confirms past literature on
the effect of atmospheric and environmental psychology onconsumer behaviour (Mehrabian and Russell, 1974; Mattila
and Wirtz, 2001). This also reiterates the importance ofproviding an atmosphere and environment that customers can
identify with and feel comfortable.
Limitation and future research directions
As with all research, ours is no exception when it comes tolimitations. We used student subjects as our sample for testing
our model. We believed that the student sample was adequateand appropriate for such a study especially in the hair care
industry. Students are familiar with the service category andhair care service is a relatively high involvement experience
that enhances the quality of the data collected. However, weshould not lose sight of the fact that due to their economic
constraints, they may have no other choice but to go to thecheapest hair stylist. This would undermine the significance
of social and relational bonds that were highlighted in thispaper. Therefore, we are cautious in generalising our
conclusions too broadly from this particular research andacknowledge the need for replication on a more representative
sample (Peterson, 2001).Also, our model was tested in a single service industry that
was considered to be very high in contact and interactionbetween employees and customers (Bowen, 1990). This may
have contributed to the positive and significant effect from thesocial/relational factors such as interaction quality and
characteristics similarity to the perceived core serviceperformance. In other service industries characterised by
less interaction and relationship building activities, thelikelihood of observing such an effect may be slimmer.
Future studies can test this in diverse service categoriesaccording to the taxonomy suggested by Bowen (1990).For future research directions, several studies have looked
at constructs such as satisfaction, commitment, trust, and
loyalty simultaneously in a single model (Garbarino andJohnson, 1999; Wang, 2002). More specifically, one of the
interests drawn from such research has been to determine
whether identical or different factors lead to satisfaction,commitment, trust, and loyalty. We see these constructs as
cornerstone concepts in marketing theory and application asthey get to the heart of what it takes to make customer
acquisition and retention successful. Although, quite often,by assuming that satisfaction, trust, and commitment are all
on identical levels of abstraction or continuum and as suchused as manifest variables to operationalise relationship
quality (Baker et al., 1999; Crosby et al., 1990; Dorsch et al.,1998; Hewett et al., 2002), they can also be perceived of as
occupying different positions on a continuum. It could bethe case that satisfaction is at the lowest level while
commitment occupies the highest with trust in the middle(White and Schneider, 2000). In such a framework, we can
imagine different antecedents to have effects on satisfaction,
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
89
trust, and commitment, which all in turn should lead to
greater loyalty. Our model takes a small step in such a
direction by confirming that when trust and loyalty are
simultaneously included in the model, relational/social
factors have an effect on trust directly and on loyalty
indirectly through trust whereas PPE affects both trust and
loyalty directly. Future research can expand our model and
others to develop an integrated framework that satisfies the
nomological validity between the above key constructs and
its antecedents.Finally, a stricter test to investigate the relationship between
the two types of service attributes using longitudinal data can
be considered. This is an important issue as the strong
relationship between the soft and hard service attributes may
be driven by common method bias-single source self reporting
data (Podsakoff and Organ, 1986). Also, we need further
studies to determine under what conditions relational/social
factors affect core service performance. This calls for a study
that examines our model across diverse service industries that
differ on the degree of interaction and contact employees have
with customers.
References
Anderson, E. and Weitz, B. (1989), “Determinants ofcontinuity in conventional industrial channel dyads”,Marketing Science, Vol. 8, pp. 310-23.
Anderson, J.C. and Gerbing, D.W. (1988), “Structuralequation modeling in practice: a review andrecommended two-step approach”, Psychological Bulletin,Vol. 103 No. 3, pp. 411-23.
Bagozzi, R.P. (1975), “Marketing as exchange”, Journal ofMarketing, Vol. 39, pp. 32-9.
Bagozzi, R.P. (1995), “Reflections on relationship marketingin consumer markets”, Journal of Academy of MarketingScience, Vol. 23, pp. 272-7.
Bagozzi, R.P. and Phillips, L.W. (1982), “Representing andtesting organizational theories: a holistic construal”,Administrative Science Quarterly, Vol. 27, pp. 459-89.
Bagozzi, R.P. and Yi, Y. (1988), “On the evaluation ofstructural equation models”, Journal of the Academy ofMarketing Science, Vol. 16, pp. 74-94.
Baker, T.L., Simpson, P.M. and Siguaw, J.A. (1999), “Theimpact of suppliers’ perceptions of reseller marketorientation on key relationship constructs”, Journal of theAcademy of Marketing Science, Vol. 27, pp. 50-7.
Baron, R.M. and Kenney, D.A. (1996), “The moderator-mediator variable distinction in social psychological research:conceptual, strategic, and statistical considerations”, Journalof Personality and Social Psychology, Vol. 51, pp. 1173-82.
Baskett, G.D. (1973), “Interview decisions as determined bycompetency and attitude similarity”, Journal of AppliedPsychology, Vol. 57, pp. 343-5.
Beatty, S.E., Mayer, M.L., Coleman, J., Reynolds, K.E. andLee, J. (1996), “Customer-sales associate retailrelationships”, Journal of Retailing, Vol. 72, pp. 223-47.
Bitner, M.J., Booms, B.H. and Tetreault, M.S. (1990), “Theservice encounters: diagnosing favorable and unfavorableincidents”, Journal of Marketing, Vol. 54, pp. 71-84.
Boles, J.S., Johnson, J.T. and Barksdale, H.C. Jr (2000),“How salespeople build quality relationships: a replicationand extension”, Journal of Business Research, Vol. 48,pp. 75-81.
Boulding, W., Kalra, A., Staelin, R. and Zeithaml, V.A.(1993), “A dynamic process model of service quality: fromexpectations to behavioral intentions”, Journal of MarketingResearch, Vol. 30, pp. 7-27.
Bowen, J. (1990), “Development of a taxonomy of services togain strategic marketing insights”, Journal of the Academy ofMarketing Science, Vol. 18, pp. 43-9.
Brady, M.K. and Cronin, J. Jr (2001), “Customer orientation:effects on customer service perceptions and outcomebehaviors”, Journal of Service Research, Vol. 3 No. 3,pp. 241-51.
Burgess, R.L. and Huston, T.L. (1979), Social Exchange inDeveloping Relationships, Academic Press, San Diego, CA.
Byrne, D.E. (1971), The Attraction Paradigm, Academic Press,New York, NY.
Cronbach, L.J. (1951), “Coefficient alpha and the internalstructure of tests”, Psychometrika, Vol. 16, pp. 297-334.
Crosby, L.A., Evans, K.R. and Cowles, D. (1990),“Relationship quality in service selling: an interpersonalinfluence perspective”, Journal of Marketing, Vol. 54,pp. 68-81.
De Wulf, K., Odekerken-Schroder, G. and Iacobucci, D.(2001), “Investment in customer relationships: a cross-country and cross-industry exploration”, Journal ofMarketing, Vol. 65, pp. 33-50.
Deutsch, M. (1962), “Cooperation and trust: sometheoretical notes”, in Jones, M.R. (Ed.), NebraskaSymposium on Motivation, University of Nebraska Press,Lincoln, NE.
Dick, A.S. and Basu, K. (1994), “Customer loyalty: towardan integrated conceptual framework”, Journal of AcademyMarketing Science, Vol. 22 No. 2, pp. 99-114.
Donavan, R. and Rossiter, J. (1982), “Store atmosphere:an environmental psychology approach”, Journal ofRetailing, Vol. 58, pp. 34-57.
Donavan, R., Rossiter, J., Marcoolynn, G. and Nesdale, A.(1994), “Store atmosphere and purchasing behavior”,Journal of Retailing, Vol. 70, pp. 283-94.
Doney, P.M. and Cannon, J.P. (1997), “An examination ofthe nature of trust in buyer-seller relationships”, Journal ofMarketing, Vol. 61, pp. 35-51.
Dorsch, M.J., Swanson, S.R. and Kelley, S.W. (1998), “Therole of relationship quality in the stratification of vendors asperceived by customers”, Journal of the Academy ofMarketing Science, Vol. 26, pp. 128-42.
Driver, C. and Johnston, R. (2001), “Understanding servicecustomers: the value of hard and soft attributes”, Journal ofService Research, Vol. 4 No. 2, pp. 130-9.
Dube, L., Chebat, J.C. and Moris, S. (1995), “The effects ofbackground music on consumers’ desire to affiliate inbuyer-seller interactions”, Psychology and Marketing, Vol. 12,pp. 305-19.
Floyd, F.J. and Wasner, G.H. (1994), “Social exchange, equity,and commitment: structural equation modeling of datingrelationships”, Journal of Family Psychology, Vol. 8, pp. 55-73.
Fornell, C. and Larcker, D.F. (1981), “Evaluating structuralequation models with unobservable variables andmeasurement error”, Journal of Marketing Research,Vol. 18, pp. 39-50.
Fornell, C., Johnson, M.D., Anderson, E.W., Cha, J. andBryant, B.E. (1996), “The American customer satisfactionindex: nature, purpose and findings”, Journal of Marketing,Vol. 60, pp. 7-18.
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
90
Frenzen, J.K. and Davis, H.L. (1990), “Purchasing behaviorin embedded markets”, Journal of Consumer Research,Vol. 17, pp. 1-12.
Frenzen, J. and Nakamoto, K. (1993), “Structure,cooperation, and the flow of market information”, Journalof Consumer Research, Vol. 20, pp. 360-75.
Ganesan, S. (1994), “Determinants of long-term orientationin buyer-seller relationships”, Journal of Marketing, Vol. 58,pp. 1-19.
Garbarino, E. and Johnson, M.S. (1999), “The different rolesof satisfaction, trust, and commitment in customerrelationships”, Journal of Marketing, Vol. 63, pp. 70-87.
Gatignon, H.A. and Robertson, T.S. (1986), “An exchangetheory model of interpersonal commmunication”, in Lutz, R.J.(Ed.), Advances in Consumer Research, Vol. 13, Association forConsumer Research, Provo, UT, pp. 534-8.
Goodman, P.S., Fichman, M., Lerch, F.J. and Snyder, P.R.(1995), “Customer-firm relationships, involvement, andcustomer satisfaction”, Academy of Management Journal,Vol. 38 No. 5, pp. 1310-24.
Goodwin, C. and Gremler, D.D. (1996), “Friendship overthe counter: how social aspects of service encountersinfluence consumer service loyalty”, in Swartz, T.A.,Bowen, D.E. and Brown, S.W. (Eds), Advances in ServicesMarketing and Management, Vol. 5, JAI, Greenwich, CT,pp. 247-82.
Gremler, D.D. and Gwinner, K.P. (2000), “Customer-employee rapport in service relationships”, Journal ofService Research, Vol. 3 No. 1, pp. 82-104.
Gronroos, C. (1984), “A service quality model and itsmarketing implications”, European Journal of Marketing,Vol. 18 No. 4, pp. 36-44.
Grove, S.J. and Fisk, R. (1997), “The impact of other customerson service experiences: a critical incident examination ofgetting along”, Journal of Retailing, Vol. 73, pp. 217-24.
Gruen, T.W., Summers, J.O. and Acito, F. (2000),“Relationship marketing activities, commitment, andmembership behaviors in professional associations”,Journal of Marketing, Vol. 64, pp. 34-49.
Gwinner, K.P., Gremler, D.D. and Bitner, M.J. (1998),“Relational benefits in services industries: the customer’sperspective”, Journal of the Academy of Marketing Science,Vol. 26, pp. 101-14.
Hayes, B.E. (1998), Measuring Customer Satisfaction:Development and Use of Questionnaires, ASQC QualityPress, Milwaukee, WI.
Hennig-Thurau, T., Gwinner, K.P. and Gremler, D.D.(2000), “Why customers build relationships withcompanies – and why not”, in Hennig-Thurau, T. andHansen, U. (Eds), Relationship Marketing: GainingCompetitive Advantage through Customer Satisfaction andCustomer Retention, Springer, Berlin, pp. 369-91.
Hennig-Thurau, T., Gwinner, K.P. and Gremler, D.D. (2002),“Understanding relationship marketing outcomes”, Journalof Service Research, Vol. 4 No. 3, pp. 230-47.
Hewett, K., Money, R.B. and Sharma, S. (2002), “Anexploration of the moderating role of buyer corporateculture in industrial buyer-seller relationships”, Journal ofthe Academy of Marketing Science, Vol. 30, pp. 229-39.
Ibarra, H. (1992), “Homophily and differential returns: sexdifferences in network structure and access in anadvertising firm”, Administrative Science Quarterly, Vol. 37,pp. 422-47.
Johnston, R. (1995), “The determinants of service quality:satisfiers and dissatisfiers”, International Journal of ServiceIndustry Management, Vol. 6 No. 5, pp. 53-71.
Jones, M.A., Mothersbaugh, D.L. and Beatty, S.E. (2002),“Why customers stay: measuring the underlying dimensionsof services switching costs and managing their differentialstrategic outcomes”, Journal of Business Research, Vol. 55,pp. 441-50.
Joshi, A.W. and Stump, R.L. (1999), “The contingent effectof specific asset investments on joint action inmanufacturer-supplier relationships: an empirical test ofthe moderating role of reciprocal asset investments,uncertainty, and trust”, Journal of the Academy ofMarketing Science, Vol. 27, pp. 291-305.
Kachra, A. and Crossan, M. (1997), Starbucks, Richard IveySchool of Business, The University of Western Ontario,London.
Kumar, N., Scheer, L.K. and Steenkamp, J.B.E.M. (1995),“The effects of perceived interdependence on dealerattitudes”, Journal of Marketing Research, Vol. 32, pp. 348-56.
Mattila, A.S. and Wirtz, J. (2001), “Congruency of scent andmusic as a driver of in-store evaluations and behavior”,Journal of Retailing, Vol. 77, pp. 273-89.
Mayer, R.C., Davis, J.H. and Schoorman, D. (1995), “Anintegrative model of organizational trust”, Academy ofManagement Review, Vol. 20 No. 3, pp. 709-34.
Mehrabian, A. and Russell, J.A. (1974), An Approach toEnvironmental Psychology, Massachusetts Institute ofTechnology, Cambridge, MA.
Meuter, L.M., Ostrom, A.L., Roundtree, R.I. and Bitner, M.J.(2000), “Self-service technologies: understanding customersatisfaction with technology-based service encounters”,Journal of Marketing, Vol. 64, pp. 50-64.
Moorman, C.R., Desphande, R. and Zaltman, G. (1993),“Factors affecting trust in market research relationships”,Journal of Marketing, Vol. 57, pp. 81-101.
Moorman, C.R., Zaltman, G. and Desphande, R. (1992),“Relationships between providers and users of marketresearch: the dynamics of trust within and betweenorganizations”, Journal of Marketing Research, Vol. 26,pp. 314-29.
Morgan, R.M. and Hunt, S.D. (1994), “The commitment-trust theory of relationship marketing”, Journal ofMarketing, Vol. 58, pp. 20-38.
Nicholson, C.Y., Compeau, L.D. and Sethi, R. (2001), “Therole of interpersonal liking in building trust in long-termchannel relationships”, Journal of the Academy of MarketingScience, Vol. 29, pp. 3-15.
O’Reilly, C.A. III, Caldwell, D.F. and Barnett, W.P. (1989),“Work group demography, social integration, and turnover”,Administrative Science Quarterly, Vol. 34, pp. 21-37.
Parasuraman, A., Berry, L.L. and Zeithaml, V.A. (1991),“Refinement and reassessment of the SERVQUAL scale”,Journal of Retailing, Vol. 67 No. 4, pp. 420-50.
Peterson, R.A. (2001), “On the use of college students insocial science research: insights from a second-order meta-analysis”, Journal of Consumer Research, Vol. 2, pp. 450-61.
Podsakoff, P.M. and Organ, D.W. (1986), “Self reports inorganizational research: problems and prospects”, Journalof Management, Vol. 12 No. 4, pp. 531-44.
Price, L.L. and Arnould, E.J. (1999), “Commercialfriendships: service provider-client relationships incontext”, Journal of Marketing, Vol. 63, pp. 38-56.
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
91
Reynolds, K.E. and Beatty, S.E. (1999), “Customer benefits andcompany consequences of customer-salesperson relationshipsin retailing”, Journal of Retailing, Vol. 75, pp. 11-32.
Smith, J.B. (1998), “Buyer-seller relationships: similarity,relationship management, and quality”, Psychology andMarketing, Vol. 15 No. 1, pp. 3-21.
Tajfel, H. and Turner, J.C. (1986), “The social identity theoryof intergroup behavior”, in Worchel, S. and Austin, W.G.(Eds), Psychology of Intergroup Relations, Nelson-Hall,Chicago, IL, pp. 7-24.
Tsui, A.S. and O’Reilly, C.A. III (1989), “Beyond simpledemographic effects: the importance of relationaldemography in superior-subordinate dyads”, Academy ofManagement Journal, Vol. 32, pp. 402-23.
Wakefield, K. and Baker, J. (1998), “Excitement at the mall:determinants and effects on shopping response”, Journal ofRetailing, Vol. 74, pp. 515-39.
Wang, G. (2002), “Attitudinal correlates of brandcommitment: an empirical study”, Journal of RelationshipMarketing, Vol. 1 No. 2, pp. 57-76.
Wathne, K.H., Biong, H. and Heide, J.B. (2001), “Choice ofsupplier in embedded markets: relationship and marketingeffects”, Journal of Marketing, Vol. 65, pp. 53-66.
White, S.S. and Schneider, B. (2000), “Climbing thecommitment ladder: the role of expectationsdisconfirmation on customers’ behavioral intentions”,Journal of Service Research, Vol. 2 No. 3, pp. 240-53.
Executive summary and implications formanagers and executives
This summary has been provided to allow managers and executivesa rapid appreciation of the content of this article. Those with aparticular interest in the topic covered may then read the article intoto to take advantage of the more comprehensive description of theresearch undertaken and its results to get the full benefits of thematerial present.
“Hard” and “soft” service attributes
The core aspects of a service are such “hard” attributes as theperformance excellence of the key service delivered. In hair
care, a “high contact” service context with significantinteraction between stylists and clients, they are the quality of
the haircut given. In contrast, the non-core aspects of a serviceare “soft” attributes. These can be based on “human” factors
such as the quality of the interaction between the stylist and the
customer resulting from the exchange of conversation, and thesimilarity of personality and tastes between the customer and
hair care provider. Soft attributes can also be non-humanfactors such the atmosphere in the salon.
The research focus
Auh explores whether social bonding between salon
employees and customers affects customers’ perceptions ofthe core service performance – the hair stylist’s ability to
provide a quality haircut. The author also examines theprocess by which social bonding attributes such as the quality
of conversation between employee and customer, and thesimilarities in the characteristics and outlook of employees
and customers, influence customers’ loyalty to the service
provider – the likelihood that a customer will return to thesalon. Finally, Auh investigates whether the process by which
these social and relational attributes affect loyalty differs fromthe way core service performance affects loyalty.
The research results
The research discovers that soft attributes positively affect thehard attribute of perceived performance excellence. Thesesoft attributes also have a positive effect on trust and anindirect effect on customer loyalty through perceivedperformance excellence and trust. In other words, the softattributes do not have a direct effect on loyalty, but only anindirect effect through perceived performance excellence andtrust. On the other hand, perceived performance excellencehas a direct effect on loyalty in addition to an indirect effecton loyalty through trust.Hair stylists who make an effort to show care and
benevolence to their customers foster the creation ofpsychological bonds that are foundations for strongrelationships. And the more the customer perceives that thestylist is striving to enhance the quality of the interaction, themore favourably the customer will perceive the performanceof the core service delivery. Similarly, when the customer andstylist share a common outlook on life, they are more likely tocommunicate well and trust each other, and the client is morelikely to believe he or she will get a decent haircut.But of course, the likelihood that the customer will visit a
particular stylist simply because of good conversation, orbecause of sharing with the stylist similar tastes and anoutlook on life, is slim. The main impetus for a client visitinga particular salon is still based on the performance of thehaircut and the trust the customer has towards the stylist inexpectation of receiving a reliable and benevolent serviceduring the client’s next visit.The research also reveals that the more readily the client
identifies with the salon’s atmosphere, the more positive he or sheis likely to be about the performance of the haircut. This isbecause the more the customer can relate to the surroundings andclimate, the more comfortable he or she will feel. This positivefeeling then carries over into his or her evaluation of the haircut.The hair care provider’s perceived ability to deliver a
reliable and quality haircut enhances the level of trust andloyalty towards the stylist. There is also a direct effect ofperceived performance excellence on loyalty, in addition to anindirect effect via trust. Trust, meanwhile, has a positive effecton customer loyalty.
How managers should react
Customers may be able to find a stylist who can do just asgood a job at providing an excellent haircut, but it may bemore difficult to find a stylist with whom they can connectand build a lasting relationship. Managers should invest inhelping employees better to interact with, and serve,customers. Managers should invest in training andeducating their employees to “click” with customers. Insituations where it takes time to learn and deliver asatisfactory core service, it may pay dividends for theorganisation to focus on the relational and social benefits toretain customers, while buying time to improve on the coreservice performance. Managers should provide an atmosphereand environment with which customers can identify, and inwhich they can feel comfortable.
(A precis of the article “The effects of soft and hard serviceattributes on loyalty: the mediating role of trust”. Supplied byMarketing Consultants for Emerald.)
The effects of soft and hard service attributes on loyalty
Seigyoung Auh
Journal of Services Marketing
Volume 19 · Number 2 · 2005 · 81–92
92