science 27

8/3/2019 Science 27

1/19

The roles of service encounters, service value, and job satisfaction in

achieving customer satisfaction in business relationships

Irene Gil , Gloria Berenguer1, Amparo Cervera 2

Faculty of Economics and Business, Department of Marketing, Avda. de los Naranjos, s/n. 46022 Valencia, Spain

Received 16 June 2006; received in revised form 4 June 2007; accepted 23 June 2007

Available online 16 August 2007

Abstract

Along with variables like the service process, perceived service value and customer satisfaction, job satisfaction of service employees plays a

vital role in customer evaluation of service result. However, there has been little in-depth research into the nature of this relation, in particular in

the context of B2B relations. In the sphere of an organization providing financial intermediation services to the banking sector and on the basis of

a literature review, hypotheses are developed which establish the mediator role of service value and the moderator role of job satisfaction of

service employees when delimiting customer satisfaction. Reliability and validity analysis give satisfactory results and our conclusions establish

firstly that service encounter directly and significantly affects perceived service value which is the final antecedent to customer satisfaction and

secondly, that the level of employment satisfaction moderates its effect on service value.

2007 Elsevier Inc. All rights reserved.

Keywords: Service encounter; Service value; Customer satisfaction; Job satisfaction of service employees; Financial sector

1. Introduction

There are fundamental differences between an organization

marketing to other organizations often referred to as industrial

or B2B marketing and an organization marketing to

consumers, that is, business to consumer (B2C) marketing

(Yanamandram & White, 2006).

The literature in general has mainly focused on consumer

services rather than business services (Parasuraman, 1998), but

driven by changes in the economy, marketing and purchasing of

business services have been receiving growing attention both inresearch and practice (Wynstra, Bjrn, & van der Valk, 2006).

Furthermore, the growth in business related services is the main

driver behind the increased share of the service sector in total

value added. In 2001, finance, insurance and business services

such as legal and consultancy services accounted for 2030% of

value added in the overall economy having doubled their

share since 1980 (Wlfl, 2005).

The study of concepts like quality, satisfaction and, more

recently, perceived value, with roots in early works by Carlzon

(1987), Grnroos (1982), Lehtinen and Lehtinen (1982),

Parasuraman, Zeithaml, and Berry (1988), and Oliver (1980),

provides new opportunities in organizational management. In particular, it becomes critical to identify and measure the

elements which contribute most to explaining satisfaction, thus

providing companies with a better understanding on how the

customer's point of view is built in an environment where

building more unique relationships with customers is vital

(Lindgreen, Palmer, Vanhamme, & Wouters, 2006). Moreover,

service marketing literature has argued that the service process,

or service encounter, may be the most important antecedent in

customer evaluation of service performance (Brown & Swartz,

1989; Lehtinen & Lehtinen, 1982). These service encounters are

considered as the basis for building customer satisfaction.

Industrial Marketing Management 37 (2008) 921939

The authors would like to express their thanks for the financial support

received under the Spanish Ministry of Education and Science Research Project

(SEJ2004-05988). Corresponding author. Tel.: +34 963 828329.

E-mail addresses: [email protected] (I. Gil), [email protected]

(G. Berenguer), [email protected] (A. Cervera).1 Tel.: +34 963 828319.2 Tel.: +34 963 828964.

0019-8501/$ - see front matter 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.indmarman.2007.06.008
mailto:irene.%E4%A7%[email protected]:[email protected]:[email protected]://dx.doi.org/10.1016/j.indmarman.2007.06.008http://dx.doi.org/10.1016/j.indmarman.2007.06.008mailto:[email protected]:[email protected]:irene.%E4%A7%[email protected]

8/3/2019 Science 27

2/19

The literature on these topics is extensive, although the

precise nature of their content and of their relationships has not

been extensively documented, even less so in terms of business-

to-business relationships (De Ruyter, Wetzels, Lemmink, &

Mattsson, 1997; Eriksson & Lfmarck Vaghult, 2000; Yeung,

Chew Ging, & Ennew, 2002).

On the other hand, in service systems it has been stated thatemployees' satisfaction with their jobs is so important as

customer satisfaction for the results of an organization (Comm

& Mathaisel, 2000:43). Employees' attitudes, in general, have

been proved as a variable affecting customer satisfaction (Adsit,

London, Crom, & Jones, 1996) and, more specifically, this

satisfaction seems to intervene in quality perceptions held by

customers (Schneider & Bowen, 1985).

With the increased demand for professional services,

marketing and organizational structures are changing and the

importance of studying the antecedents of delivering service

quality in a professional service is crucial (Boyt, Lusch, &

Naylor, 2001: 321).Particularly, Storer and Rajan (2002) point out that survival

for financial services in an evolving workplace increasingly

relies not only on technical but also on behavioral skills and

knowledge relating to working methods characterized by

networking, inter-dependency and reciprocity.

Consequently, this paper will explore relationships among

service encounter (SE), perceived service value (SV), customer

satisfaction (CS) and job satisfaction of service employees (JS),

considering as the scenario of this research an organization

specialized in service provision to financial entities, where 90%

of its activity consists of preparing and processing house

mortgages. First, we review the literature on the concepts SE,

SV, CS and JS, and identify the links between them in order todefine the research hypotheses. Then, we present the research

methodology and the results of the study, followed by the

conclusions and recommendations for management.

2. Theoretical framework and hypotheses

2.1. Service encounters

From the customer point of view, the most vivid service

impression occurs during the service encounter or moment of

truth, i.e. when customers interact with the service company

(Zeithaml & Bitner, 2002:107). During these encounters, alsoknown as interactions which take place in a relation episode

(Ravald & Grnroos, 1996), the customer receives a sort of

snapshot of the organization's level of service provision. Thus,

the result of interactions between organizations, related pro-

cesses and services, employees who provide the service and

customers define the service experience (Bitner, Faranda,

Hubbert, & Zeithaml, 1997) and from the customer's point of

view, the service encounter is the origin of the whole chain

of evaluations on the service result (Lehtinen & Lehtinen,

1982).

The service encounter has traditionally been described as the

dyadic interaction between service providers and customers

(Surprenant & Solomon, 1987). There are different types of

service encounters (Shostack, 1985), the most frequently

studied being personal interactions. Armstrong (1992) proposes

defining this process of service delivery as a system which can

be broken down into a number of different stages. Customer

perception of service characteristics in each of these stages is

therefore the antecedent and origin of any process of service

evaluation, and each encounter contributes the same to thecustomer's general satisfaction and to his/her willingness to do

business with the company again (Zeithaml & Bitner,

2002:108).

2.2. Service value

The notion of value, from a marketing approach, has a clear

subjective orientation with most authors attributing an evalu-

ative judgment to it (e.g. Berry & Yadav, 1997; Flint, Woodruff,

& Gardial, 2002; Monroe, 1992; Woodruff, 1997; Zeithaml,

1984, 1988). Furthermore, value is not inherent to services

rather it is experienced by the customers (Woodruff &Gardial, 1996:7) and therefore perceived by them. This

perception in B2B interaction materialises in judgments or

evaluations of what the customer perceives as received from the

seller in a specific situation of purchase or use (Flint et al.,

2002:103). This approach to the notion of value is consistent

with the parameters and analytical methods proposed in the

literature on consumer value (Holbrook, 1999).

There is a tendency to define value as a two-way variable

following the proposal by Oliver (1999), using the term trade-

offas equivalent to compensation or balance between benefits

and sacrifices. The most basic approach to a two-way definition

of value is that of ratio or trade-offbetween quality and price

(Monroe, 1992), in other words value for money (Fornell,Johnson, Anderson, Cha, & Bryant, 1996; Gale, 1994).

However, increasingly, authors are suggesting that this vision

is too simplistic (Bolton & Drew, 1991), and other more

sophisticated measures are needed. Thus, it is suggested that

perceived value can be understood following the proposal by

Zeithaml (1988:14), as a global evaluation that the customer

develops concerning the usefulness of a product or service,

based on the perceptions of what he or she has received in

contrast to what he or she has given. Thus, value is a positive

function of what is received and a negative function of what is

sacrificed (Oliver, 1999:45), if indeed it is possible to use the

term value to describe perceptions that are exclusively positiveor negative.

On the above basis, service value could be the result, in part,

of quality, understood as a global judgment, or attitude,

relating to the superiority of the service (Parasuraman et al.,

1988: 16). In this line of research, a significant number of

contributions present value as an advance of quality and so, it

becomes a macro-concept which includes quality (Oliver,

1999). Thus, quality components are important elements of

value although service value also includes other components

(Lapierre, Filiatrault, & Chebat, 1999:236). These other

elements would consider both the price paid for the service

and the other costs incurred by the customer on acquiring the

service.

922 I. Gil et al. / Industrial Marketing Management 37 (2008) 921939

8/3/2019 Science 27

3/19

8/3/2019 Science 27

4/19

dominant position, which establishes inverse causal order where

satisfaction is the consequence of quality.

The variables customer satisfaction and quality have also

been related to value. The concept of value, as we have shown,

goes beyond that of quality, and is an advance of quality as it

incorporates sacrifices and other additional benefits. In the

research which has observed the relationship between the two,the conclusion is that value is the consequence of quality (e.g.

Caruana, Money, & Berthon, 2000; Grewal, Monroe, &

Krishnan, 1998; Kashyap & Bojanic, 2000; Oh, 1999, 2000;

Cronin et al., 2000; Sirohi, Mclaughlin, & Wittink, 1998;

Sweeney, Soutar, Whiteley, & Johnson, 1996) and value can be

understood as a higher order construction. Referring to the link

between value and satisfaction due to the natural affinity

between the two concepts (Woodruff & Gardial, 1996:86) as

both are formed on the basis of evaluative judgments

(Woodruff, 1997), it is difficult to clearly differentiate between

them and price has arisen as the discriminant element. However,

as the number of in-depth studies on value in the literature hasgrown, the importance of price as a differentiating criteria has

started to diminish (Oliver, 1999). For Sweeney and Soutar

(2001:204), although value can easily be confused with

satisfaction, the difference is clear: these constructs are

different. While perceived value occurs in different stages of

the purchase process, including pre-purchase (Woodruff, 1997),

satisfaction is universally a post-use or post-purchase evalua-

tion. It seems clear that this statement introduces a causal order

which allows satisfaction to be understood as the result of the

perception of value, as shown in the research by, among others,

Fornell et al. (1996), Oh (1999, 2003), Caruana et al. (2000),

Babin and Kim (2001), and Gallarza and Gil (2006).

Furthermore, it could be concluded that there is a relation between the interactions which take place in what we have

called the service encounter and customer satisfaction. It could

also be stated, however, that customer judgments on these

interactions through performance scores could refer to service

value, which influences satisfaction. The issue now to be

considered concerns the nature of the relation between these

three variables. Empirical evidence is scanty and it seems that it

would be useful to confirm these relations in the sector and

sphere of activity in our research scenario.

In the service being studied, there are different interactions in

each episode which make up the service encounter. It would

seem reasonable for there to be direct effects on customerperceptions of said interactions on service value. Consequently,

the first working hypothesis we propose is as follows:

H1. The more positive the perceptions of episode interactions

the more positive the service value.

Our previous discussion hypothesized that perceptions of

service characteristics are antecedents to service value, but what

are the consequences of perceived service value? It would again

appear reasonable and in accordance with the above discussion

for perceived service value to directly affect overall customer

satisfaction raising therefore the issue of the effect of service

encounter perceptions on customer satisfaction. Thus percep-

tions of service characteristics in each interaction affect service

value which in the end affects overall customer satisfaction.

Consequently we propose the following hypotheses:

H2. The more positive the perceived service value, the more

positive overall customer satisfaction.

H3. Perceived service value mediates the effects of perceptions

of episode interactions on overall customer satisfaction.

In the competitive context of the financial services industry, it

is becoming more frequently pointed out that job satisfaction of

service employees is as important as customer satisfaction for an

organization's results (Comm & Mathaisel, 2000:43). Thus,

Schneider (1980) reports that job satisfaction is the main reason

why employees develop good service, influencing customer

satisfaction (Adsit et al., 1996;Koys, 2001; Rucci, Kim, & Quinn,

1998), their perceptions of service quality (Hartline & Ferrell,

1996; Schneider & Bowen, 1985) and competitiveness (Asif &Sargeant, 2000; Berry, 1981; Grnroos, 2001; Spinelli &

Canavos, 2000). The interactive nature of service delivery places

employees in an outstanding position to generate positive

perceptions (Zeithaml & Bitner, 2002), given that many

dimensions of service will be affected by their actions, attitudes

and emotions throughout the service encounter. In service

encounters, employees must be considered executors and

their behavioris the service quality which customers perceive

(Bitner, 1990). In this context, job satisfaction of service

employees may be understood as a motivator for service per-

formance, with this idea becoming almost an axiom in the

service literature (Wilson & Frimpong, 2004:471). Thus, and

althoughit is not the only element, Snipes et al. (2005:1330)pointout that currently most managers understand that to cause a

substantial impact on service quality in organizations, front-line

workers and customers need to be central to management con-

cerns. Introducing policies to increase job employee satisfaction

may well be worth it in the end.

Heskett, Sasser, and Schlesinger (1997) propose a theoretical

model in which employee job satisfaction begins a chain of

benefits leading to quality, productivity, service value, customer

satisfaction and loyalty which in turn leads to profits and growth.

Surprisingly, however, as Silvestro (2002) has pointed out, little

empirical evidence is given to support the validity of these

relations. The authors use the term satisfaction mirror to refer tothe fact that employee job satisfaction is reflected in terms of

customer satisfaction which in turn generates growth and profit

(Heskett et al., 1997). Similar observations are made by Heskett,

Jones, Loveman, Sasser, and Schlesinger (1994), Spinelli and

Canavos (2000), Bernhardt, Donthu, and Kennett (2000), and

Tornow and Wiley (1991). Schlesinger and Heskett (1991) and

Reichheld (1996) alsoargue that job satisfaction of employees has

the potential to improve customer service or increase customer

satisfaction. The underlying idea in these studies is that satisfied

workers will perform their work better than those who are not,

they will be in better disposition and will be more likely to behave

considerately towards colleagues and consumers (Motowidlo,

1984; Rogers, Clow, & Kash, 1994). In fact, Boshoff and Allen


8/3/2019 Science 27

5/19

(2000) conclude, among other things, that front-line employees

who are more efficient at recovering consumers are more likely to

show higher levels of job satisfaction. In the context of financial

services, Ryan, Schmit, and Johnson (1996) suggest that con-

sumer satisfaction is the cause of employee satisfaction and

Reynierse and Harker (1986) conclude that employee job sat-isfaction derives partly from the opportunity to offer customers

high service levels and partly from positive customer feedback

after the interaction. In terms of the relation employee job

satisfaction-service quality, Yoon, Beatty, and Suh (2001)

conclude that variables in job climate contribute directly to job

satisfaction and effort of service employees and indirectly impact

customer perceptions of employee service quality. The work of

Snipes et al. (2005), Hartline and Ferrell (1996), and Schneider

and Bowen (1985) is on similar lines.

Other empirical evidence, however, indicates that this rela-

tion is very weak (Ellis, Gudergan, & Johnson, 2001; Loveman,

1998; Schneider, White, & Paul, 1998; Silvestro, 2002). These

studies and other data show that the univocal relationship between employees job satisfaction and business results or

between employees job satisfaction and customer satisfaction

is beginning to be questioned, hence the appropriateness of

continuing this line of research.

Thus, we establish that moderation hypotheses can be

plausible and that the intensity of the relation between perceptions

which develop in the relation episode and service value could vary

according to the level of job satisfaction of service employees and

thus we propose the following research hypothesis:

H4. Job satisfaction of service employees moderates the effect

of perceptions in interactions in an episode on customer per-

ceived service value.

Fig. 1 is a summary of the hypotheses on the relations under

consideration.

3. Research methodology

In this work we verify the hypotheses under consideration by

analyzing the particular case of the relationship between a

company providing management services to external finance

organizations in the banking sector and their clients, banks, in

Spain. The company specializes in house mortgage service pro-

vision to financial entities, with 90% of its activity dedicated to

preparing and processing mortgages, thisservice being the coreof

our study. The request for the services provided by the external

financial organization comes directly from the credit entities(banks), which act as intermediaries between the final consumer

and the company.

The quantitative methodology uses ad-hoc interviews

through two structured questionnaires: one focused on custo-

mers (i.e. banks) and the other on the service provider

employees. The questionnaire which focuses on the customers

was administered to the managers of bank offices which, the

month before the field work began, had processed at least one

mortgage with the service provider trying to achieve a

proportion in the whole companies' customer distribution and

the sample. Using the office manager as key informant, the

questionnaire provided two-way information. Firstly, percep-

tions of the sequence of stages making up the service provisionwere evaluated, characterizing each of the interactions in the

episode or service encounter cascade on the basis of procedure

indicators (how the service is delivered) and technical and

functional (what is delivered) (Lehtinen & Lehtinen, 1982;

Ravald & Grnroos, 1996). Secondly, issues concerning the

overall evaluation of the service were considered in order to

identify service value and satisfaction. A total of 194 valid

questionnaires were obtained for analysis.

It is not common in this kind of studies to use data verifying

the randomness condition (e.g. Asif & Sargeant, 2000; Brown

& Swartz, 1989; Comm & Mahaisel, 2000; Sirohi et al., 1998;

Yoon et al., 2001). This research is not an exception. Althoughconvenience sampling presents disadvantages, for our explor-

atory research, the sample represents the kind of relationship

that the organization has with their different bank customers.

Then, given a period of time loans managed during the last

month a census of bank offices that during the period has

processed at least one loan with the company was analyzed

(194). These 194 offices were moreover proportional to the

weight of each entity in terms of the total number of loans

managed during the previous year. In this sense, in our research,

the majority of those interviewed, 63.4% of the sample,

represent bank offices of financial entity 1, with an average of

8 house mortgages; 8 house mortgages have also been handled

by financial entity 2, representing 8.3% of the total sample.

Fig. 1. Hypothesis on the relations between service encounter, service value, customer satisfaction and job satisfaction of service employees.3

3 The continuous lines represent direct relations and the discontinuous lines

represent mediated or moderated relaciones.


8/3/2019 Science 27

6/19

Finally, 28.4% of the sample are bank offices belonging to

entity 3, with 16 house mortgages handled during the last

month.

The questionnaire focused on the service provider aimed to

evaluate job satisfaction of service employees in order to

analyze its possible moderating effect on the other variables

considered. The questionnaire was administered to all the

employees in the organization. To guarantee confidentiality, it

was agreed that company managers would not be able to access

individual replies. Personal interviews were developed, during

working days. The organization under study suggested the orderof the interviews so that we would not distort the day to day

working rhythm. Questionnaires were administered by a person

hired and trained for the purpose. Data was processed and

analyzed using SPSS 11 software.

3.1. Measurement instruments

The evaluation measurements used were designed on the

basis of the literature review and through group discussions

with heads of the organization.

Our first aim is to investigate the relationships between the

provider and its organizational customer, evaluated from thislast perspective, taking, as a starting point the set of personal

interactions that take place in a service episode (Grnroos,

1990; Ravald & Grnroos, 1996) or service encounters (Bitner,

Booms, & Tetreault, 1990).

The service encounters (SE) or episodes which took place

between the provider company and the bank are direct and

indirect personal encounters. The sequence of interactions

which occurs in each of these encounters defines the service

being studied: processing a mortgage. A set of interactions have

been identified on the basis of 3 focus groups that were

composed by the organization executives and bank executives

which can be grouped around five stages which define the

service encounter: telephone service, preparation and assistance

with signing the agreement, the registering process, documen-

tation delivery and liquidation of the down payment on

expenses (see Fig. 2). Each of the five phases has been defined

on the basis of a set of service characteristics with a battery of

items (see the Appendix), evaluated through 5 possible replies

ranging from not at all appropriate (1) to very appropriate

(5), giving content to the SE scale (Service Encounter).

To evaluate service value (SV), we used a multi-item scale

which retains cost/benefit indicators in response to the need to

include multiple measurements, suggested among others by

Bolton and Drew (1991) and Sweeney and Soutar (2001). Thus,service value has been defined on the basis of quality, on the

understanding that SERVQUAL's research tradition (Parasura-

man et al., 1988) offers good opportunities as a starting point for

its evaluation. Our proposal (SV see the Appendix) is a scale

which retains one item per quality dimension and starts from

performance scores only on the lines of the research by

Brady, Cronin, and Brand (2002) and based on the recommen-

dations, among others of Novack (1997) for evaluating

organizational customers (items V1V5). After defining the

items which give content to quality, we considered including

other indicators in the original scale following the literature on

value. Thus, costs were included as effects, evaluated positively(see items V6V10 in the Appendix): (1) the perception of risk or

greater security in the experience through the indicator

confidence following the approaches, among others ofRavald

and Grnroos (1996), De Ruyter et al. (1997), and Sweeney,

Soutar, and Johnson (1999), trying to capture the emotional or

affective component of the relation's perceived value, (2) time

or effort, following the suggestion by De Ruyter et al. (1997),

defining energy or non-monetary sacrifices understood as effort,

time and convenience; (3) efficiency, considered as an element

in the value following Holbrook's proposal (1999), evaluated

through two indicators: a specific one for the organization's

staff (which Sweeney and Soutar (2001) consider is a deciding

factor in the perception of value) and another in relation to the

Fig. 2. Service encounter cascade: processing a mortgage.


8/3/2019 Science 27

7/19

Table 1

Factorial analysis suitability indicators, rotated components matrix and factor specification on the scales

SE scale Components

1 2 3 4 5

I1 0.141 7.049E02 0.141 0.116 0.880

I2 0.250 0.185 4.940E02 8.537E02 0.834

I3 0.763 6.077E

02 0.228 0.256 0.184I4 0.687 0.239 0.164 0.285 0.106

I5 0.674 0.209 0.406 5.217E02 6.896E02

I6 0.709 9.534E02 0.228 2.836E02 0.205

I7 0.579 0.379 1.85E02 0.188 0.162

I8 0.551 0.299 0.249 0.144 5.253E02

I9 0.114 0.841 0.320 0.101 0.106

I10 0.212 0.783 0.366 0.156 6.869E02

I11 0.303 0.735 0.121 0.158 0.138

I12 0.406 0.518 0.230 0.298 0.160

I13 0.372 0.185 0.625 0.223 0.225

I14 0.257 0.208 0.790 6.652E02 1.402E02

I15 0.154 0.321 0.771 8.146E02 4.741E02

I16 0.226 0.171 0.689 0.372 0.151

I17 0.203 0.117 0.172 0.820 0.204

I18 0.285 0.342 0.223 0.724 1.282E04

Factor specification F1: FACASSIST F2: FACREG F3: FACDOCDEL F4: FACDEPSET F5: FACTELSERVCronbach's alpha 0.8516 0.8676 0.8505 0.7340 0.7779

Alpha: 0.9247; correlations between variables: significant (p = zero or close to zero in almost all cases); correlation matrix determinant: 3.588E05; 2:1905.463;

degrees of freedom: 153; Bartlett's Test: 0.000; KaiserMeyerOlkin index: 0.911.

SV scale Components

1

V1 0.814

V2 0.812

V3 0.823

V4 0.785

V5 0.800

V6. 0.835V7 0.633

V8 0.704

V9 0.798

V10 0.598

Factor specification F1:SV

Alpha: 0.9089; correlations between variables: significant (p =zero or close to zero in almost all cases); correlation matrix determinant: 1.952E03; 2:1178.115;

degrees of freedom: 45; Bartlett's test: 0.000; KaiserMeyerOlkin index: 0.916.

JS scale Components

1 2 3 4 5 6

S1 6.923E02 0.180 0.144 0.839 0.151 0.110

S2 0.183 6.849E02 0.292 0.834 0.149 0.160

S3 0.763 7.966E02 8.307E02 0.245 0.255 3.45E02

S4 5.874E

02 0.149

0.191 0.404 0.623 0.383S5 0.426 0.304 0.239 8.86E02 0.421 0.143

S6 0.181 6.310E02 0.270 0.182 0.763 2.337E02

S7 0.213 9.925E03 0.325 0.101 8.699E02 0.801

S8 0.168 8.940E02 0.255 0.187 7.252E02 0.833

S9 0.306 6.585E02 0.205 0.753 0.103 4.451E02

S10 0.435 0.118 0.442 0.326 0.437 0.198

S11 0.668 5.739E02 0.169 7.099E02 0.300 0.130

S12 0.170 0.803 4.720E02 1.068E02 0.103 0.188

S13 0.336 0.148 0.480 0.197 0.593 7.53E02

S14 2.47E02 8.948E02 0.719 0.224 0.247 0.195

S15 0.109 7.070E02 0.788 0.192 0.112 0.232

S16 5.787E02 0.848 7.984E02 1.846E02 0.112 2.17E02

S17 0.176 0.182 0.817 0.105 7.86E04 0.136

S18 0.228 0.829 0.160 0.139 1.29E03 0.156

(continued on next page)


8/3/2019 Science 27

8/19

service provided; and (4) value-for-money, defined according to

the most classical proposals in the literature on value as

perceptions of the relation service quality/price, capturing thecognitive or rational component of the relation's perceived

value.

Satisfaction with the service provider (SP) was measured

overall, from a single-item scale following the conceptual

proposal by Anderson, Fornell, and Lehmann (1994). This

proposal is consistent with contributions, among others, from

Cronin and Taylor (1992), Bei and Chiao (2001), Jones and Suh

(2000), Yu and Dean (2001), and Maxham and Netemeyer

(2003), and in particular from Angur, Nataraajan, and Jahera

(1999), in the sphere of the banking industry. Overall

satisfaction was evaluated on a 10 point scale (SC).

To measure job satisfaction of service employees (JS), there

is a particularly interesting measurement proposal from Meliand Peir (1989) who designed a set of instruments (S4/82; S20/

23, S10/12 and S21/26) with notable psychometric character-

istics which could be selected according to the context in which

the questionnaire was administered. Questionnaire S20/23 with

23 items was considered appropriate (see the Appendix). The

psychometric properties of this scale were contrasted in later

studies (e.g. Gil, Berenguer, Cervera, & Moliner, 2005),

confirming their usefulness, reliability and validity.

Given that some of the scales and the items were created in

English, and they have been tested in a different language

context, to ensure the validity of the item translation, a translate/

back translate procedure (Brisles, 1970; Laroche, Papadopou-los, Heslop, & Bergeron, 2003) was used.

Finally, the classification variables such as gender, age, level

of training, length of time in the company, department or type of

contract with the organization, makes it possible to identify

employee profile.

4. Analysis of data and discussion of the results

We carried out different procedures to examine the psycho-

metric properties of the proposed measurements. The items in the

different scales were analyzed on the basis of procedures recom-

mended in the methodology on scale design for evaluating

marketing constructs (Churchill, 1979; Diamantopoulos &

Winklhofer, 2001; Judd & McClelland, 1998), and, where

appropriate, of applying principal components analysis (herein-

after referred to as PCA) and confirmatory factor analysis (CFA).Reliability analysis, understood as internal consistency, was

carried out by calculating Cronbach's coefficient and using

measurements which relate each isolated item (item-total cor-

relation and inter-item correlation).

In the Service Encounter (SE) scale, the statistics analyzed

confirm scale properties, and the items appear to show correct

behavior and the measurements and typical deviations for each

item appear approximately equal. Correlations of the indicators

with respect to the total are moderate or high and positive in all

cases. Finally the coefficient reaches a value of 0.9247, which

is a good result, showing score stability and internal consis-

tency. The structure of the relations between the variables in the

scale was verified by PCA with orthogonal rotation using theVARIMAX method. The application of this statistical technique

to our data, was supported by different criteria based on the

correlation matrix. Both the KMO value (above 0.9) and

Bartlett's test of Sphericity (p below the critical level of 0.01),

indicate it is appropriate to develop a PCA. The results show

there are five factors which coincide with the 5 stages defined,

explaining 70.504% of the variance (see Table 1).

The first factor which emerges can be considered as the

preparation/assistance with the signature phase, it accounts

for 19.15% of the variance and groups six indicators on

assistance with preparing the signature, staff training, analysis

of register viability, mistakes in preparing the signature,information prior to the signature and speed of any modifica-

tion. The second and third factors describe respectively the

registration/processing process and the documentation deliv-

ery phase and account for 15% of the variance. The contents of

the four indicators in the second factor are related to the

information, time used and the response to any event during

deed processing. The contents of the indicators for the third

factor relate to correct documentation, the form and method of

delivering documentation and the time elapsing between the

date it is dispatched and received. The fourth factor includes

two items which explain 9% of the variance and refer to the

study for the coverage of funds and the settlement sheet and

has been termed settling the deposit. Finally we can identify

JS scale

JS scale Components

1 2 3 4 5 6

S19 0.691 0.195 0.267 7.337E02 0.230 0.329

S20 0.662 0.202 2.74E02 0.432 0.170 0.204

S21 0.363 0.627 9.670E02 0.176 3.34E02 1.118E02

S22 0.723 0.247

2.26E

02 0.121 0.345 7.020E

02S23 0.100 0.709 1.113E02 9.922E02 8.638E02 0.268

Factor

specification

SAP: satisfaction

with provision

SAPE: satisfaction with

physical working

environment

SAPAR:

participation

satisfaction

IS: intrinsic work

satisfaction

SIR: satisfaction

with inter-personal

relations

SASU: satisfaction

with supervision

Cronbach's

alpha

0.8744 0.8504 0.8238 0.8744 0.8073 0.8682

Alpha: 0.9155; Raju's : 0.8684; correlations between variables: significant (p =zero or close to zero in almost all cases); correlation matrix determinant: 1.150E07;

2: 998.646; degrees of freedom: 253; Bartlett's test: 0.000; KaiserMeyerOlkin index: 0.800. Those items in bold are the ones loading higher on each factor. An item

was considered to load on a given factor if the factor loading obtained in the rotated factor matrix was 0.4 or greater.

Table 1 (continued)


8/3/2019 Science 27

9/19

Table 2

CFA results of measures

CFA results of measures for JS scale

Items (JS) R2 Reliability Factor

S3 0.59 0.35 Composed rel iability = 0.8504 AVE = 0. 5358 SAP: satisfaction with provision

S11 0.66 (5.54) 0.46

S19 0.75 (5.85

) 0.56S20 0.84 (6.21) 0.71

S22 0.76 (5.67) 0.63

S12 0.63 0.40 Composed reliability=0.8524 AVE=0.5440 SAPE: satisfaction with physical working environment

S16 0.76 (7.76) 0.57

S18 0.95 (4.99) 0.89

S21 0.74 (4.34) 0.54

S23 0.56 (4.97) 0.31

S14 0.75 0.57 Composed rel iability = 0.8286 AVE = 0. 6178 SAPAR: participation satisfaction

S15 0.85 (6.15) 0.71

S17 0.76 (5.28) 0.57

S1 0.8 0.64 Composed reliability = 0.8751 AVE = 0.7014 IS: intrinsic work satisfaction

S2 0.92 (9.57) 0.85

S9 0.79 (9.39) 0.62

S4 0.45 0.30 Composed reliability=0.7568 AVE=0.5265 SIR: satisfaction with inter-personal relations

S6 0.74 (3.10

) 0.55S13 0.91 (2.92) 0.82

S7 0.86 0.81 Composed rel iability = 0.8428 AVE = 0. 7284 SASU: satisfaction with supervision

S8 0.90 (7.60) 0.62

Discriminant validity Correlation2

SAPSAPE = 0.2570 SAPESASU=0.0936

SAPSAPAR = 0.1739 SAPAR IS=0.3169

SAPIS = 0.3226 SAPAR SIR=0.3844

SAPSIR = 0.4045 SAPAR SASU=0.3306

SAPSASU = 0.2285 ISSIR=0.3114

SAPESAPAR = 0.1163 ISSASU=0.2116

SAPEIS = 0.1102 SIR SASU=0.1459

SAPESIR=0.1399

Chi-squared (174)=206.43 p =0.04; GFI=0.806; CFI=0.947; RMSEA=0.053; BB-NNFI=0.935; BB-NFI=0.795

CFA results of measures for SE scale

Items (SE) R2 Reliability Factor

I3 0.79 0.62 Composed reliability = 0.8561 AVE = 0.5003 FACASSISTI4 0.77 (10.86) 0.56

I5 0.75 (10.38) 0.56

I6 0.67 (9.13) 0.47

I7 0.61 (7.36) 0.39

I8 0.63 (7.83) 0.40

I9 0.72 0.52 Composed reliability = 0.8398 AVE = 0.5675 FACREGI10 0.78 (17.69) 0.59

I11 0.75 (9.98) 0.56

I12 0.78 (9.92) 0.60

I13 0.77 0.59 Composed reliability = 0.8499 AVE = 0.5863 FACDOCDELI14 0.75 (12.07) 0.56

I15 0.76 (11.12) 0.60

I16 0.79 (10.34) 0.59

I17 0.69 0.47 Composed reliability = 0.7422 AVE = 0.5929 FACDEPSETI18 0.85 (8.21) 0.716

I1 0.73 0.53 Composed reliability = 0.7845 AVE = 0.6472 FACTELSERVI2 0.87 (6.79) 0.76

Discriminant validity Correlation2

FACASSISTFACREG= 0.4651 FACREGFACDEPSET=0.5595

FACASSISTFACDOCDEL= 0.4277 FACREGFACTELSERV=0.2285

FACASSISTFACDEPSET= 0.3819 FACDOCDELFACDEPSET=0.4624

FACASSISTFACTELSERV= 0.2873 FACDOCDELFACTELSERV=0.1616

FACREGFACDOCDEL= 0.4583 FACDEPSETFACTELSERV=0.1475

Chi-squared (125)=174.03 p =0.002; GFI=0.897; CFI=0.969; RMSEA=0.046; BB-NNFI=0.961; BB-NFI=0.900

929I. Gil et al. / Industrial Marketing Management 37 (2008) 921939

8/3/2019 Science 27

10/19

telephone service which accounts for 9% of the variance, a

value very close to factor four. As shown in Table 1, the

coefficient in the factors varies between 0.73 and 0.86. After

defining the factors and studying their internal consistency, we

proceed to analyze the validity of the proposed scale. In ouropinion, the measurement verifies content validity as the

domain of the construct has been specified and all the possible

dimensions and contents for the concept under analysis have

been considered. On the same lines, we can also state the

nomological validity of the scales using Pearson's correlations

between scale items and the overall evaluations for each of the

five stages (item-test correlation) through IGi indicators.

As the Service Value (SV) scale was defined as additive, four

basic aspects were considered (Hair, Anderson, Tatham, & Black,

1999): (1) conceptual definition; (2) dimensionality; (3) reliabil-

ity; and (4) validity. In terms of conceptual definition, in our

proposal there is a correspondence between the individual items

and the concept, as shown in the literature review. In relation todimensionality, the scale shows a unidimensional behavior

achieving an explained variance of 58.417 (see Table 1). The

third underlying assumption is reliability, in this sense, the relation

statistics of each item with the other items behave correctly. Item

correlation with respect to the total is high in all cases. The

coefficient shows the test's high internal consistency, with a value

of 0.9089. Finallyin relationto scale validity, the scale shows high

criteria or concurrent validity as the correlation matrix between

items and a single-item criterion external to the test which we call

GV (Global Value) were all positive and neither very high nor

very low. Correlation of the resulting sum scale with the item GV

makes it possible to confirm concurrent validity of the SV scale.Finally, nomological validity, observedby correlating SV with the

SE scale is verified for both scales.

In relation to customer satisfaction, the correlation coeffi-

cient was calculated between items for the evaluation of each

stage and the overall evaluation, with significant correlations in

all cases.

Finally, the multi-item scales of job satisfaction of service

employees (JS) used were defined as additive, and checked for

concept, reliability, dimensionality and validity (Hair et al.,

1999:105106). The job satisfaction of service employees JS

scale was found to be satisfactory in terms of reliability.

Relational statistics showed correct behavior for all the items,

with moderate to high, positive correlations in all cases with

respect to the total. Internal scale consistency measured by

Cronbach's alpha was 0.9155 which is very close to the 0.92

obtained by Meli and Peir. A principal components analysis

(PCA) was performed to analyze scale components with

orthogonal rotation using the VARIMAX method. Applicationof this statistical technique to our data was supported by

correlation matrix based criteria.

As Table 1 shows, the 23 items were grouped into six factors

explaining 72.560% of the total variance, using latent root

criteria. The first component which clearly emerges relates five

items explaining 15.047% of the variance and is known as

satisfaction with provision SAP. The second factor which

can be termed satisfaction with the physical working environ-

ment SAPE, accounts for 11.588% of the variance explained,

grouping five items. The third factor, participation satisfac-

tion SAPAR, includes three items which explain 7.707% of

the variance. The fourth factor accounts for 6.552% of the

variance and has been called intrinsic work satisfaction IS.Two final factors emerge, the fifth factor which accounts for

5.5% of the variance and a sixth factor accounting for 5.2% of

the variance, both have two items each, denominated respec-

tively as satisfaction with inter-personal relations SIR, and

satisfaction with supervision SASU. The internal consisten-

cy of the set of items used to evaluate each type of job

satisfaction of service employees, defines an alpha in the factors

which oscillates between 0.80 and 0.87. If the small number of

items in each factor is taken into account, reliability can be

considered excellent. For each factor, correlations between the

different indicators item-total were greater than 0.5 in all cases.

Once the exploratory factor structure of the scales wasdelimited, we proceeded to confirm the obtained factors through

confirmatory factor analysis (CFA). The measurement model

estimation was performed with the Robust Maximum Likeli-

hood Method using the asymptotic variancecovariance matrix

and with the EQS 6.1 software package.

We firstly calculated the internal consistency of the

dimensions that compose each scale, considering simultaneous-

ly two indicators: composed reliability coefficient whose

minimum threshold value is 0.7 (Anderson & Gerbing, 1988;

Bagozzi & Yi, 1988), and the extracted variance in each

scalewhose value must exceed 0.5 (Fornell & Larker, 1981).

These indexes, all of them shown in Table 2, proved to be

acceptable for all factors obtained.

Discriminant validity

CFA results of measures for SV scale

Items (SV) R2 Reliability Factor

V1 0.76 0.57 Composed reliability = 0.9182 AVE = 0.5323 SV

V2 0.80 (12.65) 0.63

V3 0.80 (11.28) 0.64

V4 0.78 (9.98

) 0.61V5 0.75 (10.02) 0.57

V6 0.66 (9.76) 0.43

V7 0.56 (7.18) 0.31

V8 0.78 (17.87) 0.61

V9 0.59 (9.33) 0.34

V10 0.78 (15.10) 0.61

Chi-squared (35)=49.78 p =0.031; GFI=0.939; CFI=0.983; RMSEA=0.051; BB-NNFI=0.977; BB-NFI=0.952. Note: loadings are significative at 99%.

Table 2 (continued)


8/3/2019 Science 27

11/19

Taking into account the global result obtained, we can state

that there is a sufficient level of internal consistency in the

items, what means that the ability of the set of items employed

to represent each of the latent constructs is satisfactory.

Finally, construct validity (convergent and discriminant

validity) of the scales was analyzed. In the case of the unidi-

mensional scale SV, convergent validity can be stated as all itsvariables are associated to significant and high loadings at least at

95% (tN1.96). In the case of the multi-dimensional scales, (SE

and JS scales), convergent validity was corroborated given that,

apart from the significant loadings obtained, correlations among

different dimensions loading on a second latent factor proved to

be significant at 95%. Therefore, it can be stated that scales have

convergent validity (Anderson & Gerbing, 1988).

Discriminant validity, which means that each factor

represents a separate dimension, was analyzed through lineal

correlations or standardised covariances among latent factors.

These values showed evidence of discriminant validity given

that they showed values well below the unit. Furthermore, it

was verified that extracted variance in each of the dimensionswas higher than square correlations (Table 2). It can be stated

therefore that constructs show discriminant validity (Fornell &

Larcker, 1981).

After the above analysis it can be concluded that factors are

stable, corroborating the exploratory structure, and dimensions

show validity, so they can therefore be retained for theory testing.

Given the global results obtained on the quality of the mea-

sures, we can state that the subscales employed in our ques-

tionnaire are validated.

At a global level, other indexes for corroborating the

goodness of fit of the model provided satisfactory results

(Chi-squared; RMSEAb0.08; GFI close to or higher than 0.9;Bentley Bonnet normed and non-normed (1990) indexes BB-

NFI, BB-NNFI close to or higher than 0.9; and Compared Fix

Index CFIN0.9 for all subscales).

After analyzing measurement accuracy, regression analysis

with mediation was used to contrast the hypotheses on the links

between the constructs analyzed. Regression analysis has

sometimes been used to evaluate the predictive capacity of

service quality measurements (Angur et al., 1999) or to researchthe influence of perceived price and service quality on satisfaction

(Mittal & Lassar, 1996; Novack, 1997) and on loyalty (Bei &

Chiao, 2001; Lee & Cunningham, 2001; Zeithaml, Berrry, &

Parasuraman, 1996), to determine the moderating role of value in

satisfaction (Caruana et al., 2000) and that of satisfaction on

loyalty (De Ruyter & Bloemer, 1999).

Modelling including mediation processes is common in

social psychology, other previous research focused on variables

in relation to satisfaction have recently applied this methodol-

ogy with success (e.g. Bei & Chiao, 2001; Gil et al., 2005;

Maxham & Netemeyer, 2003). In general, a variable can be

defined as a mediator to the extent that it influences the relation

between the predictor and the criterion. When the measurementmodel involves latent constructs, modelling through structural

equations provides the basis for the data analysis strategy. If the

measurement model only involves measured variables, the basic

analytical approach is multiple regression (Kenny, Kashy, &

Bolger, 1998). Whatever data analysis method is used, the steps

required to test mediation are the same. In accordance with

Baron and Kenny (1986:1177), three regression equations

should be estimated to contrast mediation: first the mediator is

regressed on the independent variable, second, the dependent

variable is regressed on the independent one and, third, the

dependent variable is regressed on both the independent vari-

able and the mediator. These three equations are the basis forobserving relations in a mediation model under the following

Table 4

Analysis of the association between interactions in the episode and customer satisfaction

Regression equation 2: SESC

Independent

variables

F1:

FACASSIST

F2:

FACREG

F3:

FACDOCDEL

F4:

FACDEPSET

F5:

FACTELSERV

Constant R R2 R2

corrected

Standard

error

F model Durbin

Watson

0.274 0.335 0.206 0.330 0.173 5.585

(177.582)0.675 0 .455 0 .441 0. 6733 31. 411 1.834

(5.657) (6.917) (4.252) (6.809) (3.573)

0.305 0.372 0.229 0.367 0.192

Significance less than or equal to 0.01; significance less than or equal to 0.05 but greater than 0.01; significance less than or equal to 0.10 but greater than 0.05.

Student's absolute t statistic value is shown in brackets under the respective estimated parameter which represents the non-standardised coefficient. The value whichappears below the t statistic is the standardised coefficient.

Table 3

Analysis of the association between interactions in the episode and service value

Regression equation 1: SESV

Independent

variables

F1:

FACASSIST

F2:

FACREG

F3:

FACDOCDEL

F4:

FACDEPSET

F5:

FACTELSERV

Constant R R2 R2

corrected

Standard

error

F model Durbin

Watson

2.468 1.765 1.628 1.617 0.839 44.146

(242.062)

0.841 0.707 0.699 2.54019 90.780 1.814

(13.498) (9.653) (8.906) (8.842) (4.587)0.533 0.381 0.352 0.349 0.181


Student's absolute t statistic value is shown in brackets under the respective estimated parameter which represents the non-standardised coefficient. The value which

appears below the t statistic is the standardised coefficient.

931I. Gil et al. / Industrial Marketing Management 37 (2008) 921939

8/3/2019 Science 27

12/19

conditions: first, the independent variable must affect the

mediator in the first equation; second the independent variable

must affect the dependent variable in the second equation; and

third, the mediator must affect the dependent variable in the

third equation. If all the conditions are verified in the predicted

direction, then the effect of the independent variable on the

dependent one must be less in the third equation than in thesecond. Perfect mediation occurs if the independent variable

loses its influence when the mediator is included in the equation

(Kenny et al., 1998:260). In this case, the influence of the

independent variable on the dependent variable disappears

completely in the presence of the supposedly mediating variable

(Chumpitaz & Vanhamme, 2003:81).

Thus, to examine whether the proposed hypotheses are

fulfilled and the mediating effects associated with the sequence

SESVCS, first we regressed the variables related to

service value (mediator), to the predictive variables related to

service encounter. Thus, the regression 1 equation in Table 3

analyses the effect of the independent variable service encounter

on the mediators in service value. As can be seen, serviceencounter dimensions significantly and positively affect service

value. Thus, the first mediation condition has been satisfied for

most of the variables, verifying H1.

Secondly, we regress service encounter dimensions (predic-

tive variables) to overall satisfaction. The regression 2 equation

in Table 4 shows that service encounter dimensions signifi-

cantly and positively affect overall satisfaction. The second

mediation condition can therefore be said to be fulfilled, as the

predictive variable affects the dependent variable.

The third and fourth mediation conditions are observed in the

same regression equation, as the dependent variable is regressed

on both the mediator and the predictors. As in the above

sections, we carry out multiple linear regression analysis. The

first variable to enter is service value (SV) followed by factor 4

and factor 2 from the scale interactions in the episode, excluding

the other factors from the model as they are not significant. For

the entire mediation to be statistically supported, the effects of

the predictive variable in the dependent variable must not besignificant, dominated by the mediator. However, as we have

already indicated, complete mediation is not very common in

the social sciences (Baron & Kenny, 1986; Maxham &

Netemeyer, 2003) and in this context it is considered that

partial mediations can support a mediation hypothesis (Kenny

et al., 1998). Thus when the effects of the predictive variables in

the dependent variable decrease in equation 3 (dominated by the

mediator) with regard to those obtained in equation 2 (not

dominated by the mediator), mediation is supported (see Table

5). In regression equation 3, the mediator is significant and in

addition the beta coefficients for the predictive variables (the

factors which define interactions 2 and 4 which have been

significant) are smaller than regression equation 2 (its valuedecreases from 0.367 to 0.171 in FACPROCEL and from 0.372 to

0.159 in FACASSIST) and its significance decreases from 0.01 to

0.05 (Liu, Luo, & Shi, 2002). Thus the effect of the interactions

on satisfaction diminish in size and significance when service

value is controlled in the regression. Service value is shown as a

mediator which significantly decreases the relation between the

dependent and independent variable rather than eliminating it.

Finally, in the last model, tolerance, i.e. the proportion of

variance in each predictive variable not explained by the other

predictive variables takes values close to 0.8 and the inverse, the

variance inflation factor (VIF), takes values close to 1.2 which

Table 6

Values measured by department in the measures of job satisfaction which show different perceptions

Department Job satisfaction (JB) SAP: satisfaction

with provision

SAPAR:

participation

satisfaction

IS: intrinsic work

satisfaction

SIR: satisfaction

with inter-personal

relations

Average Standard

deviation

Average Standard

deviation

Average Standard

deviation

Average Standard

deviation

Average Standard

deviation

Reception 3.97 0.49 3.84 0.40 3.46 0.83 4.00 0.61 4.50 0.50

Authorization 3.18 0.48 2.61 0.67 2.95 0.71 3.43 0.78 3.67 0.99

Signings 3.63 0.55 2.72 0.96 3.53 1.06 4.25 0.63 4.25 0.82

Tax liquidation 2.97 0.49 2.42 0.88 2.66 0.56 3.25 0.85 3.06 0.62

External relations 3.74 0.64 3.44 0.88 3.76 0.78 3.65 0.81 3.80 0.58

Verification and control of documentation 3.33 0.44 2.75 0.83 3.54 0.61 3.31 0.65 4.00 0.75

Table 5

Analysis of the association between interactions in the episode, service value and customer satisfaction


Independent

variables

SV F1:

FACASSIST

F2:

FACREG

F3:

FACDOCDEL

F4:

FACDEPSET

F5:

FACTELSERV

Constant R R2 R2

corrected

Standard

error

F model Durbin

Watson

0.109 n.s. 0.143 n.s. 0.154 n.s. 3.769

(7.332)

0.710 0.504 0.496 0.6391 64.338 1.905

(9.404) (2.834) (3.093)0.561 0.159 0.171





8/3/2019 Science 27

13/19

denotes little co-linearity with no redundant or superfluousvariable.

Overall, the results support the idea that service value par-

tially mediates the links between SE and SC. Thus, it can be

seen that: (1) variations in the perceptions of interactions

significantly influence variations in the levels of perceived

service value. Institutional customer perceptions of the technical

and functional characteristics of the encounter cascade therefore

do directly affect service value; (2) variations in the perceptions

of the interactions significantly influence variations in satisfac-

tion levels, with a partially mediated effect through service

value and (3) variations in service value significantly influence

satisfaction. Thus including service value in the model

intensifies its power of explanation as it supplies informationon how and why the interactions affect satisfaction. Thus, par-

tial mediation has been supported and H2H3 confirmed.

After studying the behavior of the mediating variable, wethen studied the effect of the service encounter on service value

and customer satisfaction to see if it was modified according to

the level of job satisfaction of service employees.

The variable job satisfaction of service employees shows

differences both overall and by satisfaction components, except

for intrinsic satisfaction and satisfaction with the physical

environment, according to the department the employee belongs

to. Table 6 shows the basic descriptives for these variables. The

highest averages in these departments are observed in satis-

faction with relations and the lowest for satisfaction with

benefits and participation. The tax liquidation department

shows the lowest levels of satisfaction while overall the most

satisfied employees are in reception and signing. All theseresults show that employee perceptions differ according to the

department where they work and that means we can focus on the

Table 7

Analysis of the associations between interactions in the episode and service value according to the level of job satisfaction of service employees

Independent

variables

F1:

FACASSIST

F2:

FACREG

F3:

FACDOCDEL

F4:

FACDEPSET

F5:

FACTELSERV

Constant R R2 R2

corrected

Standard

error

F model Durbin

Watson

JS HIGH

N=127

2.721 1.229 1.605 1.263 0.657 44.181

(193.081)0.834 0.695 0.683 2.4998 55.175 1.608

(11.145) (5.290) (6.687) (5.282) (2.902)

0.571 0.270 0.337 0.266 0.147JS LOW

N= 67

2.449 2.549 1.603 2.008 1.052 43.941

(141.413)0.889 0.791 0.774 2.3589 46.188 2.376

(8.645) (8.987) (6.103) (7.592) (3.588)

0.524 0.540 0.358 0.450 0.211

TOTAL

SAMPLE

N=194

2.468 1.765 1.628 1.617 0.839 44.146

(242.062)0.841 0.707 0.699 2.54019 90.78 1.814

(13.498) (9.653) (8.906) (8.842) (4.587)

0.533 0.381 0.352 0.349 0.181

Difference

analysis

t=0.7758 t=3.7058 t=0.0057 t=2.1325 t=1.0915




Table 8

Analysis of the association between interactions in the episode, service value and customer satisfaction in situations of high and low job satisfaction of service

employees


Independent

variables

F1:

FACASSIST

F2:

FACREG

F3:

FACDOCDEL

F4:

FACDEPSET

F5:

FACTELSERV

SV Constant R R2 R2

corrected

Standard

error

F model Durbin

Watson

JS HIGH

N=127

0.324 0.440 0.218 0.323 0.187 8.541

(107.634)

0.767 0.538 0.550 0.6025 17.433 2.009

(4.481) (6.074) (3.251) (4.780) (2.497)0.381 0.513 0.268 0.397 0.206

JS LOW

N= 67

0.251 0.282 0.198 0.334 10.170 8.610

(131.889)0.629 0.396 0.371 0.7132 15.858 2.032

(3.602) (4.259) (2.897) (4.898) (2.626)

0.260 0.306 0.206 0.347 0.187


JS HIGH

N=127

n.s. 0.137 n.s. 0.195 n.s. 0.108 3.828

(5.714)0.682 0.465 0.452 0.6656 35.648 2.032

(2.110) (2.894) (7.174)

0.149 0.202 0.532

JS LOW

N= 67

n.s. n.s. n.s. n.s. n.s. 0.135 2.625

(3.955)0.741 0.550 0.543 0.6105 79.298 1.730

(8.905)

0.741


Student's absolute t statistic value is shown in brackets under the respective estimated parameter which represents the non-standardised coefficient. The value whichappears below the t statistic is the standardised coefficient.


8/3/2019 Science 27

14/19

perceptions of employees in the authorization department, the

one with the maximum responsibility for providing the service

being studied. Thus for each customer request there is an

employee in that department who monitors the different stages

of the service and acts as contact employee, in other words as

the service's front-line employee for the customer. In this

context and in the light of the literature review, we understandthat the attitudes expressed by these employees could influence

customer perception of the value of the service provided. Thus

we went on to examine whether job satisfaction of service

employees is a moderating variable for the relation defined

between the interactions in the episode and service value. In this

way, we established dyadic relations between these employees

and the banks involved in the analysis.

There are different analytical procedures for testing modera-

tion. Moderation implies that the causal relation between two

variables changes according to the moderating variable. The

statistical procedure should therefore measure and test the

differential effect of the dependent variable on the independent

oneas a function of the moderator. The channelfor comparing and

measuring this effect depends in part on the measurement levelfor

the independent variable and the moderating variable. In our

research scenario, we considered it appropriate to distinguish

between employees with the highest satisfaction scores and those

with the lowest, defining two groups of employees: those above

average and those below average. Thus in overall job satisfaction

of service employees we differentiate two levels and, followingBaron and Kenny (1986), we measure the effect of the

independent variable on the dependent variable through non-

standardised regression coefficients. We then test the difference

between these coefficients(see Cohen & Cohen, 1983). Theresult

was a group of 127 banks which work with the employees we

have named as more satisfied and a second group of 67 banks

working with those less satisfied. After identifying the groups

the objective is to observe whether the associations between the

predictive variables and the criterion are significantly different.

Table 7 shows the relations between the variables for the two

levels of job satisfaction of service employees identified by

regression analysis. Following Sharma and Patterson (2000) wefirst contrasted the two groups using Chow's test.

Chow's test shows that the forms and slopes of the two

regression models are significantly different, the value of the

statistic F=3,2534NF(6,182) at 0.05 means that H0 on

structural stability can be rejected while also verifying that the

dependent variable service value is explained in a significantly

different way in both models. Chow's test evaluates whether

there are global differences in intra-group parameter values but

it does not evaluate the significance of the individual estimated

parameters. The statistically significant differences between the

individual regression coefficients were analyzed using tests

recommended by Cohen and Cohen (1983). The analyses used

to investigate the significance of the differences in the definedregression coefficients in the two groups (see the last row in

Table 7) identify differences in the regression coefficients

associated to the perceptions of the bank in the stages identified

by process of registering/processing and settlement of the

deposit. Observation of the corresponding estimated para-

meters shows that both interactions have a lower impact on

service value in the banks with more satisfied employees than in

those with less satisfied employees, thus confirming the

moderating character of job satisfaction of service employees

and consequently H4. Note that these factors were the ones

which showed direct effects on overall customer satisfaction,

which suggests that in the two groups, service value contributesdifferent explanations for overall customer satisfaction. A

regression analysis with mediation was then considered for

both groups. Table 8 shows the last regression model for the two

mediation assumptions, and it can be seen that for the group

with low job satisfaction, service value completely mediates the

effect of interactions on overall satisfaction while in the

situation of high job satisfaction there are direct effects on

factors 2 and 4 in global satisfaction. This second sequence is

identical to the behavior observed overall.

The global statistics for each of the groups (Table 9) in the

variables in scales SE and SV, show that banks with more

satisfied contact employees perceive slightly higher levels of

service value in all the indicators, with higher perceptions of

Table 9

Analysis of average differences in the groups

JS

high

JS

low

I1. Waiting time before the switchboard takes your call 4.34 4.36

I2. Waiting time before being able to speak to the appropriate

person

4.14 4.30

IG1. Overall evaluation of the telephone service 4.40 4.52

I3. Service provided during preparation of the signing 4.36 4.17

I4. Preparation and technical training of the staff 4.53 4.29

I5. Pre-authorization or legal report with analysis of register

viability

4.33 4.15

I6. Mistakes or errors in preparing the signing 4. 39 4.31

I7. Information provided before the signing 4. 28 4.23

I8. Speed of attending to any last minute change or

modification

4.37 4.30

IG2. Overall evaluation of in-company service 4.24 4.17

I9. Information provided during processing the deeds 4.23 4.15

I10. Information provided in the operations processed 4.20 4.18

I11. Time used in the processing phase 3.96 3.86

I12. Response to any event 4.38 4.36

IG3.Overall evaluation of behavior in the processing phase 4.23 4.19I13. Accuracy in the documents we deliver 4.37 4.45

I14. Form of delivering documentation (packaging, order, etc.) 4.41 4.45

I15. Method of delivering documentation 4.28 4.36

I16. Period between the date of dispatch and reception 4.14 4.20

IG4.Overall evaluation of documentation delivery 4.31 4.41

I17. Coverage of funds study 4.10 4.09

I18. Settlement sheet 4.30 4.02

IG5.Overall evaluation of the liquidation phase 4.29 4.08

V1. Reliability 4.62 4.51

V2. Professionalism 4.63 4.57

V3. Shows interest 4.52 4.49

V4. Rapid response 4.47 4.43

V5. State of the art management services 4.29 4.27

V6. Trust 4.64 4.47

V7. Problem solving 4.39 4.26V8. Efficient service provision 4.39 4.32

V9. Efficient staff 4.57 4.52

V10. Quality/price relation 3.95 3.69

Significance less than or equal to 0.05.


8/3/2019 Science 27

15/19

service value and in general more positive perceptions with

regard to the interactions in the episode.

A t test on the averages for the technical and functional

characteristics which describe each service stage shows that the

service provided during the preparation of the signing, the

report and the preparation and training for staff providing the

service to the bank are significantly different in the two groups,with a lower value in the group of employees with a lower level

of job satisfaction.

Thus, we can state that job satisfaction of service employees

moderates the effect perceptions on the interactions in an episode

have on service value, such that in situations of low job satisfaction

the effect of the predictors process of registering/processing and

settlement of deposit on the service value intensifies, with service

value acting here as interveningvariable. While in situations of high

job satisfaction of service employees the interactions occurring in

the relation episode affect overall customer satisfaction thus

generating proof that high employee job satisfaction permits

encounters with the bank that contribute to explaining satisfaction.

5. Conclusions and future lines of research

For more than two decades, scientific research has dealt with

the issues analyzed in this work the service encounter,

service value, customer satisfaction and job satisfaction of

service employees and some of the relations between them,

occasionally in the scenario of B2B relations, with fruitful but

not conclusive results (Yeung et al., 2002). We consider that this

is therefore an opportunity to develop research which will help

to further this line of study and at the same time respond to

Parasuraman's (1998) call for research.

In fact, as pointed out in the introduction to this paper, theliterature in general has shown a predominant focus on consumer

services rather thanbusiness services, but fundamental differences

between B2B and B2C marketing invite for specific research in

this area. Moreover, changes in the economy, with the growing

importance of both business related services and professional

services gives rise to specifically developing research in this area.

From a conceptual point of view, the contributions of this

study relate to the definition conceptualization and oper-

ationalization of the constructs analyzed and empirical

analysis of the relation between all the proposed concepts.

We consider that the definition and operationalization of

perceived value is of particular interest. The literature review hasestablished the basic proposition that value can be defined on the

basis of quality, incorporating another type of benefits and

sacrifices. Furthermore, in the sphere of inter-organizational

relations, the origin of the evaluations of this perceived value are

to be found in the interactions between the service provider and

the customer in each episode of the relation. The development of

a measurement scale for this construct adapted to the context

under study (Eriksson & Lofmark Vaghult, 2000) and, in this

case, a B2B relation may be an interesting contribution.

In the sphere of the relation between the concepts of value,

customer satisfaction and job satisfaction of service employees

throughout the service encounter, verification of the hypotheses

has led to several conclusions.

Considering the service encounter as a dyadic provider

customer interaction has provided a suitable framework for

analyzing the contributions of perceptions from both collectives

to service evaluation. Thus in the sphere of intermediation

services for a mortgage, the stages in the service encounter have

been defined and service value proved to be a variable which

partially mediates the effect of the interactions in each episodeon overall customer satisfaction. Perceptions concerning the

encounter cascade generate direct, mediated effects through

service value, on institutional customer satisfaction. Thus, the

evaluation process originates in the interactions which take place

in the relation episode and this suggests that this process can be

better understood by including service value as a mediating

variable in these perceptions and customer satisfaction.

Furthermore, considering the encounter as dyadic underlines

the determining role of the employee in service provision. The

literature suggested the mirror of satisfaction concept to

describe the relation between job satisfaction of service employ-

ees and customer satisfaction, establishing that satisfied employ-ees perform their work better and contribute to increasing

customer satisfaction levels. Thus the employee has a decisive

role in service provision. If employees are part of a solid service

culture and receive management support for delivering improved

service to the client, this more positive experience may lead to

increased job satisfaction of service employees and thus influence

customer perceptions. Our hypotheses proposed analyzing

whether employee job satisfaction moderated the existingrelation

between the perceptions which occur in the relation episode and

service value. Empirical evidence found in the literature sug-

gested that this relation is very weak and the results were not very

conclusive and even the reverse and therefore analysis of the

proposed relations was a research opportunity.

The empirical results indicate that the level of employee job

satisfaction for those in charge of monitoring the service being

studied appears to moderate the relation between customer per-

ceptions in episode interactions and service value. We can

thereforesuggest that a way of addingservice value and increasing

customer satisfaction is through employee job satisfaction.

Finally, despite the fact that the objectives considered have

been verified, this work presents a series of limitations which

open lines for future research. Firstly, the use of just one

company for the research invites a repeat study in other

companies in the sector and in other service contexts in order

to confirm or reject the relations found here. Replica studies arenecessary to validate the scales used with larger, random

samples. The non-randomness of the process for selecting the

units for research and their limited size, generate reservations on

generalizing the results. Secondly, cross-cultural approaches to

the constructs analyzed would be interesting to check for

differences in conceptualization and operationalization. Thirdly,

the line of research opened provides the opportunity to observe

how the constructs under investigation evolve with time. It may

well be relevant to use longitudinal approaches for a better

understanding of the dynamic behavior of the variables

analyzed. Fourth, future research needs to be done on the links

between variables analyzed and the expected relationships

which turned out to be weaker.


8/3/2019 Science 27

16/19

One important element for further research would be related to

theimportance of emotion andof the affective side in perceptionsof

relationships among providers and customers in industrial settings,

as previously analyzed in bank customers by Barnes (1997).

All these aspects underline the importance of front-line

employees in the creation of value and satisfaction for customers

as also proved in industrial settings, and, therefore, the criticalnature of human resources management for companies, not only

in consumer settings, but also in industrial spheres.

Finally we should emphasize the need to try out new

analytical methods to enrich and confirm the present results,

both with regard to measurement capacity and the relations

identified among them all. The problem of measurement in

marketing has been approached with two main types of indexes

(Diamantopoulos & Winklhofer, 2001): reflective and forma-

tive. This study was developed based on formative indexes

which, because they are aggregates of simple items means there

are no measurement errors and they can be incorporated directly

into explanatory techniques such as the regression model. Theycannot, however, recognize the existence of latent factors. A

future line of research would therefore be to study alternative

methodologies in-depth which make it possible to unite the

capacities of both types of index such as for example specific

developments of LISREL to incorporate formative indexes with

latent variables. These developments, based on MIMIC models

are very interesting and are a line of research to be followed. The

problem with them and the reason why they were difficult to

apply in this study is that a large sample size is needed to provide

a robust approach to the multiple relations which emerge.

Appendix A. Measurement scales in the model

SE. Service encounter

Customer perceptions of technical and functional characteristics of the service

provided by the provider organization, reflecting the service encounter

cascade or interactions in a relation episode

I1. Waiting time before the switchboard takes your call

I2. Waiting time before being able to speak to the appropriate person

I3. Service provided during preparation of the signing

I4. Preparation and technical training of the staff

I5. Pre-authorization or legal report with analysis of register viability

I6. Mistakes or errors in preparing the signing

I7. Information provided before the signing

I8. Speed of attending to any last minute change or modification

I9. Information provided during processing the deedsI10. Information provided in the operations processed

I11. Time used in the processing phase

I12. Response to any event

I13. Accuracy in the documents delivered

I14. Form of delivering documentation (packaging, order, etc.)

I15. Method of delivering documentation

I16. Period between the date of dispatch and reception

I17. Coverage of funds study

I18. Settlement sheet

SV. Service value

The advantages perceived in exchange for the charges borne (Berry and Yadav,

1997: 29)

Judgments or evaluations of what the customer perceives he has received from

the seller in a specific purchase or use situation (Flint et al., 2002: 103)

V1. The service provider is reliable

V2. The service provider is professional

V3. The service provider shows interest

V4. The service provider responds quickly

V5. The service provider has state of the art equipment and infrastructures and

equipment in management services

V6. I trust the service provider

V7. The service provider solves problems for me

V8. Efficiency of the service provided

V9. The service provider's staff are efficient

V10. Quality/price relation of the service they provide

CS. Overall client satisfaction/IGi. Satisfaction by stages of service delivery

This is a global evaluation based on total consumption experience ( Anderson

et al. 1994: 54; Fornell, 1992: 11).

A global measurement of a set of satisfactions with specific previous

experiences (Yu & Dean, 2001: 235)

CS. Overall client satisfaction

IG1. Telephone service

IG2. In company service

IG3. Behavior in the processing phase

IG4. Documentation delivery

IG5. The liquidation phase

JS. Job satisfaction of service employees

Positive, pleasurable emotional state resulting from the evaluation of one's own

work or of one's employment experiences (Locke, 1969)

S20/23 (Meli & Peir, 1989)

S1. The job itself (in general).

S2. The opportunity your job offers to do the things you do well

S3. The salary you receive

S4. The objectives. goals you must reach

S5. The training opportunities offered by the company

S6. Personal relations with your superiors

S7. Supervision of your work

S8. The proximity and frequency of that supervision

S9. The chance to do things you enjoy in your job

S10. The way your supervisors (or superiors) judge your work

S11. Equality and Fairness in the way you are treated by the company

S12. Physical environment and the space you have in your workplace

S13. The support you receive from your superiors

S14. The capacity for autonomous decisions on aspects of your job

S15. Participation in the decisions made by your department or section

S16. Lighting in your workplace

S17. Participation in working group decisions concerning the company

S18. Ventilation in your workplace

S19. Opportunities for Promotion

S20. The extent to which the company complies with the collective agreement.

labor law and regulations

S21. The temperature in the workplace

S22. How negotiation on employment aspects is carried out in the company

S23. Cleanliness and hygiene in the workplace.

References

Adsit, D. J., London, M., Crom, S., & Jones, D. (1996). Relationships between

employee attitudes, customer satisfaction and departmental performance.

Journal of Management Development, 15(1), 6275.

Angur, M. G., Nataraajan, R., & Jahera, J. S. (1999). Service quality in the

banking industry: An assessment in a developing economy. International

Journal of Bank Marketing, 17(3), 116123.

Anderson, E. W., Fornell, C., & Lehmann, D. R. (1994, July). Customer

satisfaction, market share, and profitability: Findings from Sweden. Journal

of Marketing, 58, 5366.

Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in

practice: A review and recommended two-step approach.Psychological

Bulletin, 103(3), 411423.


8/3/2019 Science 27

17/19

Armstrong, P. K. (1992). Analyzing quality in the service delivery process.

Service productivity and quality challenge conference (pp. 2324). The

Wharton School.

Asif, S., & Sargeant, A. (2000). Modeling internal communications in the

financial services sector. European Journal of Marketing, 34(3/4), 299317.

Babin, B. J., & Kim, K. (2001). International students travel behavior: A model

of the travel-related consumer/dissatisfaction process. Journal of Travel and

Tourism Marketing, 10(1), 93

106.Bagozzi, R. P., & Yi, Y. (1988). On the evaluation of structural equation models.

Journal of the Academy of Marketing Science, 16(2), 7494.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in

socialpsychological research:Conceptual,strategic, andstatisticalconsiderations.

Journal of Personality and Social Psychology, 51(6), 11731182.

Barnes, J. G. (1997).

science 27

Documents