Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

International Journal of

,i ! R Research in
...... (w Marketing
ELSEVIER Intern. J. of Research in Marketing 13 (1996) 29-51

Consumer evaluations of new technology-based self-service


options: An investigation of alternative models of service quality
Pratibha A. Dabholkar *
University of Tennessee, Department of Marketing and Logistics / Transportation, 307, Stokely Management Center, Knoxville, TN
37996, USA
Received 15 July 1994; accepted 15 June 1995

Abstract

Increasing labor costs and advances in technology are encouraging service firms to consider offering technology-
based self-service options to consumers. Yet, little is known about how consumers would evaluate such options. Two
alternative models of service quality are proposed based on an attribute versus overall affect approach. The attribute
model is based specifically on what consumers would expect from such options. The overall affect model is based on
the consumer's feelings toward the use of technology. A study is conducted to test these models for a new (proposed)
technology-based self-service option, with and without the effect of different waiting times. Implications for firms in
terms of service design and promotion are discussed, and directions for future research are indicated.

Keywords: Consumer evaluation; Service delivery; Technology-based self-service; New technology-based service;
Self-service options; Service quality; Models of service quality; Attribute-based decision model; Affect-based
decision model; Effect of waiting time

I. Introduction Banking, 1986). 1 T h e benefits of self-service de-


livery options are evident in terms of productivity
High labor costs are causing service firms to and cost savings for the firm (Chase, 1978; Love-
examine delivery options that allow consumers to lock and Young, 1979; Mills and Moberg, 1982;
p e r f o r m services for themselves (Business Week, Schneider and Bowen, 1985). Yet, little is known
1990). Moreover, advances in technology are about c o n s u m e r p r e f e r e n c e for such options, par-
making self-service innovations (e.g., a u t o m a t e d ticularly those based on technology. Popular
car rental machines at airports, c o m p u t e r order- sources suggest that consumers are willing to
ing in stores) increasingly feasible (World of engage in self-service if they perceive it as m o r e

1 A comprehensive discussion on current and potential


* Tel.: (615) 974-1656; fax: (615) 974-1932; e-mail: technology-based self-service options and the implications of
pa146028@utkvml.bitnet. these for service marketers is presented in Dabholkar (1994a).

0167-8116/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved


SSDI 0167-8116(95)00027-5
30 P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51

convenient, efficient, or enjoyable (The New York important research issues-service quality and cus-
Times, 1989) and that they are becoming increas- tomer participation in technology-based service
ingly comfortable with technology (Chain Store delivery. The study represents one of the first
Age Executive, 1987). However, little systematic attempts to apply consumer decision making con-
research has been done to investigate customer cepts to the services area and explores an-
evaluations of service quality for technology-based tecedents of service quality in terms of decision-
self-service options. making theory. In addition, these proposed mod-
Service quality is a critical issue for both con- els are tested under different situational condi-
sumers and service providers. Consumers seek tions in order to investigate the impact of situa-
and demand high quality services (Sherden, 1988; tional factors on service quality and customer
The Wall Street Journal, 1990a), while service participation in service delivery. The results of
providers hope that quality services will enhance the study suggest strategic implications for firms
their image, sales, and profitability (Buzzell and in terms of designing and promoting self-service
Gale, 1987; Berry et al., 1989; Gummesson, 1993). innovations based on technology. Directions for
Although service quality has received much atten- continued research on customer evaluations of
tion in the literature (Gronroos, 1984; Jacoby and service quality, customer participation issues, and
Olson, 1985; Parasuraman et al., 1988) and the technology-based service delivery are suggested.
concept of 'customer participation,' i.e., cus-
tomers actively participating in service delivery,
has been discussed (Bateson, 1985; Schneider and 2. Conceptual framework
Bowen, 1985; Silpakit and Fisk, 1985), no concep-
tual framework has been developed to study ser- 2.1. Developing alternative models of service qual-
vice quality for self-service options in general or ity
technology-based self-service options in particu-
lar. Much of the literature on 'customer participa-
In an exploratory study, Langeard et al. (1981) tion' in service delivery (i.e., self-service options)
examined factors relevant for self-service (in gen- is based on a management perspective of quality.
eral) and found among other things that time and Customers are viewed as 'partial employees'
control were important to those who preferred whose participation or performance in service
self-service options. Ledingham (1984) explored delivery can be used by the service firm to im-
the use of videotext services at home and found prove the quality of its operations (e.g., Schneider
that time savings were important to consumers and Bowen, 1985; Mills and Morris, 1986; Kelley
but could not compensate for the loss of social et al., 1990). Customer participation in this paper
interaction. Cowles and Crosby (1990) investi- is viewed from the customer's perspective in or-
gated the 'personalness' of new communication der to investigate the factors underlying the cus-
media and found that people have different toler- tomer's decision to 'participate' in the service
ances for replacing people with machines. Al- delivery. Similarly, service quality in this paper is
though these studies have not proposed a concep- based on the customer's perspective which often
tual framework to study self-service or technol- differs from the producer's evaluation and is criti-
ogy-based self-service options, nor have they in- cal for predicting shopping behavior (Zeithaml,
vestigated service quality with respect to such 1988; Gronroos, 1993). Jacoby and Olson (1985)
options, they have laid the groundwork for pro- suggest that subjective quality (or customer evalu-
ceeding with such research. ation) is more closely tied to sales performance,
The purpose of this study is to propose and and therefore merits more attention than objec-
test alternative models of service quality for tech- tive quality (or management evaluation).
nology-based self-service options by drawing on Given that technology-based self-service op-
consumer decision-making research. A concep- tions are relatively new forms of service delivery,
tual framework is developed to integrate two service firms need strategic direction regarding
P.A. Dabholkar / lntern. J. of Research in Marketing 13 (1996)29-51 31

whether or not to offer such options and how to these options and thus would not be able to
design and promote them. As of now they have indicate their perceptions of reliability and other
little knowledge about decision-making models dimensions. For the same reason, recently sug-
that consumers would use to evaluate and select gested service quality models based only on per-
technology-based self-service options. Traditional ceptions (Cronin and Taylor, 1992) would also not
service quality models based on disconfirmation, be applicable. The only types of evaluations cus-
i.e., expectations-perceptions (Gronroos, 1983; tomers would have, about service delivery that
Parasuraman et al., 1988) would be difficult to they have not yet tried, would be expectations.
apply as consumers may not have experienced Further, technology-based self-service options

ATTRIBUTE-BASED MODEL
(Cognitive Evaluation of Characteristics Associated with
Technology-Based Self Service Options)

Expected Speed of D e l i v e r ~ H!

Expected Ease of Use ~ _ ~ . . " Expected Intention


Service Quality of H8 to Use
H3
Expected Reliability ~ Technology-Based Technology-Based
Self-Service Option Self-Service Option
Expected EnjoyH m e n ~ 5 / / ~ "H' 4~ '
/ -
Expected Controlf

OVERALL AFFECT MODEL


(Affective Process of Evaluation based on Disposition toward
Technology-Based Self-Service Options)

Attitude toward Using.


Technological Products~H6

~",,~, Expected Intention


Service Quality of H8 to Use
Technology-Based Technology-Based
Self-Service Option Self-Service Option

Need for I n t e r a c t i o n / ~
with Service Employee

Fig. 1. Alternative models of service quality for technology-basedself-serviceoptions.


32 P.A. Dabholkar / Intern. Z of Research in Marketing 13 (1996) 29-51

r e p r e s e n t a u n i q u e form of service delivery, a n d D r a w i n g o n c o n s u m e r d e c i s i o n - m a k i n g re-


the d i m e n s i o n s of service quality suggested in search, two a l t e r n a t i v e m o d e l s of service quality
t r a d i t i o n a l m o d e l s m a y n o t apply. Interviews with are proposed: a n a t t r i b u t e - b a s e d m o d e l a n d a n
p o t e n t i a l c u s t o m e r s of t e c h n o l o g y - b a s e d self- overall affect m o d e l (see Fig. 1). T h e attribute-
service o p t i o n s a n d a n e x a m i n a t i o n of past stud- based m o d e l is b a s e d o n a cognitive a p p r o a c h to
ies o n service delivery, self-service, a n d the use of decision making, w h e r e c o n s u m e r s would use a
technological p r o d u c t s m u s t suggest the a p p r o p r i - c o m p e n s a t o r y process ( J o h n s o n , 1984; D a b -
ate d i m e n s i o n s of service quality. holkar, 1994b) to evaluate a t t r i b u t e s associated

Table 1
Determining relevant factors for evaluating service quality of technology-based self-service options
Factors relevant to Findings from Past research findings
customers in using qualitative research on service delivery,
technology-based specificallyconducted self-service, and
self-service options for this study use of technological products
Characteristics associated with technology-based self-service options:
Time taken to use Yes Lovelock and Young, 1979
Langeard et al., 1981
Hornik, 1984
Ledingham, 1984
Maister, 1985
Effort required to use Yes Langeard et al., 1981
Davis et al., 1989
Bagozzi, 1990 (ease of use)
Complexity of process Yes Williams et al., 1985
Cowles, 1989 (personalness)
Reliability of outcome Yes Evans and Brown, 1988
Parasuraman et al., 1988
Van Gorder, 1990
Accuracy of outcome Yes Davis et al., 1989, Davis et al., 1992,
Bagozzi, 1990 (performance)
Enjoyment in using option Yes Langeard et al., 1981
Holbrook et al., 1984
Davis et al., 1992
Holbrook and Hirschman, 1982 (novelty)
Control over process No Langer and Saegert, 1977
Langeard et al., 1981
Bowen, 1986
Bateson and Hui, 1987
Guiry, 1992
Disposition toward technology-based self-service options:
Attitude toward using technological products Yes Dickerson and Gentry, 1983
Ledingham, 1984
Williams et al., 1985
Need for interaction with service employee Yes Langeard et al., 1981
Ledingham, 1984
Breakweli et al., 1986
Cowles, 1989
Forman and Sriram, 1991
P.A. Dabholkar/ Intern. J. of Research in Marketing 13 (1996) 29-51 33

with the technology-based self-service option in of technology-based self-service options to deter-


order to form expectations of service quality. mine the attributes they considered important
Some consumers may have a predetermined idea with respect to such options. Integrating their
about salient attributes, and they may review responses, it was found that time, effort, complex-
these cognitions to form evaluations of service ity, reliability, accuracy, and enjoyment were im-
quality. Other consumers may have little prior portant to customers in evaluating and deciding
knowledge about the service delivery, but would to use technology-based self-service options. The
think about the process, form cognitions about interviewees were also asked what else might
relevant attributes, and then combine them to influence their evaluations of such options, and
evaluate service quality. Although this process most people indicated that the way they felt about
may be somewhat complicated, compensatory using technology and about interacting with ser-
evaluations are much more likely in unfamiliar vice employees would be relevant. The findings of
situations (Bettman and Park, 1980; Park and this qualitative research are compared with re-
Lessig, 1981) and where there are few alterna- sults from a number of past studies on service
tives facing the customer (Lussier and Olshavsky, delivery, self-service, and the use of technological
1979). products (see Table 1). These two sources to-
The overall affect model is based on an affec- gether suggest the variables that are incorporated
tive approach to decision making where con- in the alternative models of service quality in Fig.
sumers would use overall predispositions (John- 1. The hypotheses underlying these models are
son, 1984; Dabholkar, 1994b) to form expecta- discussed below. 2
tions of service quality for a specific technology-
based self-service option. These predispositions 2.2. Attribute based model
may be toward using technological products in
general or toward using self-service in general. As seen from Fig. 1, five attributes of service
This model would be more likely if consumers delivery are important to potential customers of
already had strong predispositions (Bettman and technology-based self-service options: speed of
Park, 1980) toward technology a n d / o r self- delivery, ease of use (effort and complexity), reli-
service, if they wanted to minimize the effort ability (includes accuracy), enjoyment, and con-
involved in forming evaluations (Johnson, 1984), trol.
or if they were naturally inclined to make deci-
sions based on affect rather than cognitions. 2.2.1. Speed o f delivery
For both models, expected service quality Langeard et al. (1981) found that time was
would influence intentions to use the technol- very important to those who preferred self-service.
ogy-based self-service option (see Fig. 1). Al- Similarly, I_~dingham (1984) found that time sav-
though little empirical research has been done to ings were important to people in using electronic
support the direct link between service quality banking and shopping. Time in these studies could
and intentions to use the service, service firms be interpreted to include waiting time as well as
assume that providing quality services will in- the time taken for active delivery of the service.
crease sales (Buzzell and Gale, 1987; Berry et al., The results of the qualitative research also sug-
1989), and researchers suggest that it is a logical gested that time was important to potential cus-
relationship which should be modeled and tested tomers of technology-based self-service options,
in various service situations (Jacoby and Olson, but respondents specifically mentioned both
1985; Zeithaml, 1988). speed of service delivery and waiting time. In this
The relevant attributes and dispositions re-
garding technology-based self-service options are
2 The modelsdeveloped in this paper are equally applica-
uncovered from past studies and from qualitative ble to technology-basedself-serviceoptionsthat the customer
research specifically designed for this study. In- has tried, simply by substituting 'perceptions' for 'expecta-
terviews were conducted with potential customers tions' in the consumer'sevaluationof the service delivery.
34 P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51

paper, waiting time is examined separately as a Customers may be concerned about ease of
situational factor and expected speed of delivery use for several reasons. One reason may be re-
is defined as the customer's expectation of the lated to saving actual effort expended. Another
time it would take to actively perform the service. reason may be to reduce social risk. If customers
The rationale for separating these two factors is expect the technology to be difficult to use, they
that one may view the actual use of an ATM as may become concerned about social risk (i.e.,
fast, but there may be a long waiting line (e.g., they may fear looking foolish as they struggle to
Kolesar, 1984). use it) and will view this as a low quality option.
Customers today are highly sensitive to the Instead, if they expect the technology to be easy
speed of service delivery (The Wall Street Jour- to use, they will view the service delivery based
nal, 1990b; Bateson, 1985; Silpakit and Fisk, on this technology as a high quality option.
1985), and studies have shown that they usually
overestimate the time taken to deliver a service H2: Expected ease of using the technology-
(Hornik, 1984). Lovelock and Young (1979) sug- based self-service option will have a positive ef-
gest that some people prefer to perform the fect on expected service quality.
service themselves, specifically to reduce delivery
time. This may be explained by Maister's (1985) 2.2.3. Reliability
observation that unoccupied time feels longer Parasuraman et al. (1988) have identified relia-
than occupied time. Maister (1985) also suggests bility of the service as an important determinant
that slow service delivery affects our overall per- of service quality. A recent survey of service
ceptions of service quality. Thus, if customers customers (Van Gorder, 1990) found reliability to
expect that a service will be delivered speedily, be the most important factor in determining ser-
they are likely to evaluate the service more highly. vice quality. Studies on the use of computer tech-
nology have also found performance or depend-
HI: Expected speed of delivery of the technol-
ability of the options to be an important dimen-
ogy-based self-service option will have a positive
sion (Davis et al., 1989; Bagozzi, 1990; Davis et
effect on expected service quality.
al., 1992). Customers may be especially con-
2. 2. 2. Ease o f use cerned about the reliability of new service deliv-
Langeard et al. (1981) found that in deciding ery options based on technology because they
between alternative service delivery options, cus- may envision some performance risk in that these
tomers considered the effort involved in using the options may not work well (Evans and Brown,
service delivery to be somewhat important. The 1988). Finally, some respondents interviewed in
qualitative research suggested that some poten- the qualitative study mentioned that reliability
tial customers of technology-based self-service and accuracy of the outcome would be relevant
options were concerned about the effort required for evaluating technology-based self-service op-
to use such options and the complexity of the tions.
process of service delivery. These two character-
istics - effort and complexity -appear to be re- H3: Expected reliability of the technology-
lated and encompassed in 'ease of use,' found to based self-service option will have a positive ef-
be an important attribute to customers in using fect on expected service quality.
computer technology (Davis et al., 1989; Bagozzi,
1990). Some researchers have referred to the 2. 2. 4. E n j o y m e n t
'personalness' of different media (Williams et al., Respondents in the qualitative investigation
1985) and others to the 'personalness' of auto- said that they would be more likely to use a
mated service options (Cowles, 1989) in referring technology-based self-service option if it looked
to the responsiveness or ease of use of these like something that would be enjoyable. Langeard
options and found that it was an important fac- et al. (1981) found that some people enjoy playing
tor. with machines and suggest that these people may
P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51 35

prefer self-service options that allow them to do as actual control; moreover, it can increase the
so. Similarly, Holbrook et al. (1984) point out the evaluation of the experience (Langer and Saegert,
enjoyment aspect of computer software and 1977). Bateson and Hui (1987) propose that by
games, and Holbrook and Hirschman (1982) sug- increasing perceived control one can enhance the
gest that the novelty aspect encourages customers value of the service to the customer. Applying the
to try new products. Research on the use of same logic, it is proposed that by increasing ex-
computer technology (Davis et al., 1992) found pected control, the expected value of the service
that customers value the fun involved in using to the customer would be enhanced.
such products. Thus, for technology-based self-
service options, we expect enjoyment (arising in- H5: Expected control in using the technology-
trinsically from interacting with such options or based self-service option will have a positive ef-
from the novelty aspect) to be import'ant to cus- fect on expected service quality.
tomers in evaluating such options.
2.3. Overall affect model
H4: Expected enjoyment from using the tech-
nology-based self-service option will have a posi-
As seen from Fig. 1, two general dispositions
tive effect on expected service quality.
are important to potential customers of technol-
ogy-based self-service options: attitude toward us-
2.2.5. Control
ing technological products and need for interac-
In the context of a service encounter, per-
tion with the service employee.
ceived control has been described as the amount
of control that a customer feels h e / s h e has over
the process or outcome (Bateson and Hui, 1987; 2.3.1. Attitude toward using technological products
Langeard et al., 1981). Extending the concept Many of the technology-based self-service op-
from perceptions to expectations, expected con- tions are fairly new (e.g., automated airline ticket-
trol is defined as the amount of control a cus- ing machines) and the majority of customers may
tomer expects to have over the process or out- not be familiar with them. In such cases, people
come of a service encounter. Langeard et al. tend to make judgments about the new situation
(1981) found that control is important to cus- based on related past experiences and attitudes.
tomers in using self-service. Bateson (1985) and For example, in communications research, the
Bowen (1986) suggest that people choose self- user views new communications options as mere
service options not for monetary savings, but to extensions of existing technologies (Williams et
feel in control. Guiry (1992) suggests that control al., 1985). Most people today are likely to have
would be important to customers in any type of been exposed to some technological products
service delivery. In a comparison with the find- (e.g., computers, ATMs, VCRs) and to have
ings of past research, control was the only vari- formed an attitude toward using such products,
able not uncovered in the qualitative research ranging anywhere from very favorable to very
specifically conducted for this study (possibly be- unfavorable. This type of generalized attitude is
cause it is a factor people may find difficult to referred to as category-based affect (Fiske, 1982).
articulate). Nevertheless, control is likely to be Empirical studies (Dickerson and Gentry, 1983;
quite relevant for evaluating technology-based Ledingham, 1984) have shown the existence of
self-service options, given that it is a factor intrin- such generalized attitude and its influence on the
sically important to most people. evaluation of new, but similar, situations.
A person's belief that h e / s h e has (or will
have) control, even in the absence of real control, H6: Attitude toward using technological prod-
will result in benefits similar to those associated ucts (in general) will have a positive effect on the
with real control (Glass and Singer, 1972; Langer, expected quality of the technology-based self-
1975). This sense of control can be as significant service option.
36 P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51

2.3.2. Need for interaction with service employee lieves that a service delivery option will be high in
Although Langeard et al. (1981) failed to find quality would intend to use that option, as long as
that the need for interaction with a person influ- price and other situational factors are the same
enced the selection of self-service options, they for alternative service delivery options.
believed this to be a relevant factor for such
contexts. For many service encounters, human H8: Expected service quality of the technol-
interaction is extremely important in evaluating ogy-based self-service option will have a positive
the service (Solomon et al., 1985; Bitner et al., influence on intention to use that option.
1990). Especially for services where the customer
is present, customers evaluate the quality of the 2.5. Effect of situational factors
process by the nature of the interaction
(Gronroos, 1982; Surprenant and Solomon, 1987). Silpakit and Fisk (1985) indicate that situa-
Some people feel strongly that the use of ma- tional factors should have a strong influence on
chines in a service encounter dehumanizes the the use of self-service. Situational conditions
interaction (Breakwell et al., 1986; Zeithaml and could include crowding, whether the customer is
Gilly, 1987). Ledingham (1984) found that the alone, with friends or family, or in a hurry. Situa-
loss of social interaction in a self-service situation tional factors may become evident only after ar-
could not compensate for time savings. On the riving at the service site, such as waiting time for
other hand, some people enjoy playing with ma- a given service delivery option. In general, one
chines (Langeard et al., 1981) or computer games would expect that waiting time would adversely
(Holbrook et al., 1984) and this may reduce their impact intentions to use a service delivery option,
need for interaction with service employees. Stud- especially if the customer is in a hurry. With
ies show that people have different perceptions regard to the other situational factors, the out-
of automated technologies based on the impor- come is not as clear. For example, if people are
tance of retail contact to them (Forman and unfamiliar with a technology-based self-service
Sriram, 1991) and as a result have different toler- option or uncomfortable with technology in gen-
ances for replacing people with machines in ser- eral, they may be less likely to use the option in
vice encounters (Cowles, 1989; Cowles and crowded conditions. On the other hand, people
Crosby, 1990). who are uncomfortable speaking to (or in front
of) others may be more likely to use the technol-
H7: Need for interaction with the service em- ogy-based self-service option under crowding. 3
ployee will have a negative effect on the expected Similarly, it is possible that customers would be
quality of the technology-based self-service op- more likely to use a technology-based self-service
tion. when alone, just to have something to do (Mais-
ter, 1985). On the other hand, some customers
2.4. Service quality-intention link may be less likely to seek the technology-based
self-service option if alone because they need
The next hypothesis fits in with both models human interaction (Forman and Sriram, 1991).
shown in Fig. 1. From the service finn's perspec-
tive, unless expected service quality is translated
3 Although the literature does not indicate how customer
into actionable behavior, there is little benefit for would evaluate technology-based self-service options under
the firm. Service providers work on the assump- crowded conditions, aspects of crowding were examined in
tion that providing quality services will lead to this study in an exploratory sense. Crowding was manipulated
greater use of their service (Jacoby and Olson, as a treatment and manipulation checks showed that condi-
1985; Buzzell and Gale, 1987; Zeithaml, 1988; tions of crowding were noted by respondents. However,
crowding was found to have no effect on service quality or
Berry et al., 1989). Moreover, the link between intentions to use technology-based self-service options. For
expected quality and intentions to use a given the sake of brevity and due to the exploratory nature of these
service option is intuitive. An individual who be- hypotheses, crowding is not discussed in this paper.
P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51 37

Given that some of these situational conditions 3.1. Research context and design
may work differently for different individuals,
only waiting time is examined in this study. The research hypotheses were tested (using a
Waiting time in service situations has been scenario and questionnaire approach) for a con-
studied empirically by applying queuing theory to text where customers can either use a computer-
service industries with waiting lines such as tele- ized, touch screen to order a meal in a fast food
phone service and airlines (Czepiel, 1980). Wait- restaurant (technology-based self-service) or or-
ing time is also relevant for many service situa- der verbally. This particular context was selected
tions such as fast-food restaurants, banking, gro- for several reasons. First, touch screens in a fast
cery stores, department stores, discount stores, food setting are currently being tested as direct
post offices, and public health delivery systems. ordering options for customers (Marketing News,
Chebat and Filiatrault (1993) studied the impact 1990; Nation's Restaurant News, 1991). Second,
of videos at the service site on reducing perceived fast food restaurants are frequented by a more
waiting time. The rationale for such studies is representative sample of the general population
that customers today place enormous emphasis than are airports, catalog stores, movie theaters,
on time and simply hate to wait in line (The New or sports events (The Scarborough Market Re-
York Times, 1989; The Wall Street Journal, port, 1986). Third, the touch screen technology is
1990b). Given a choice, people will tend to choose easier to use than, for example, computer key-
the option with less waiting time (Clemmer and boards and would be easier to visualize or think
Schneider, 1989; Dabholkar, 1990; Langeard et about. Finally, the situation itself involves a lower
al., 1981; Maister, 1985). Further, waiting time is level of risk than one where the outcome is
likely to adversely affect evaluations of service critical (e.g., using a medical innovation to moni-
quality (Dube-Rioux et al., 1989; Taylor, 1994). tor one's own health). Thus, the fast food setting
Thus: is a good choice for initial theory development
and for suggesting relevant implications for prac-
titioners. Research implications could extend to
H10a: Waiting time will have a negative effect
other on-site, service encounters which offer simi-
on intention to use the technology-based self-
lar technology-based self-service options.
service option.
It was decided to use a scenario approach over
a field study for several reasons. In testing a new
H10b: Waiting time will have a negative effect and as of yet widely unavailable technology-based
on expected service quality of the technology- self-service option, a study of potential customers
based self-service option. and their expectations was thought to be appro-
priate. Secondly, given the comprehensive models
being tested, the questionnaire was too long to be
administered in a field study to customers of
3. Method 'fast' food. Responses in that setting would not
be as thoughtful, and furthermore, less hurried
The primary research objectives of this study people would self-select themselves into the sam-
were to test alternative models of service quality ple, creating non-respondent bias. In addition,
(cognitive versus affective) for technology-based using the scenario approach allowed manipula-
self-service options, to determine which attributes tion of waiting time, something not easily repli-
a n d / o r dispositions are important in evaluating cated in field experiments (Mixon, 1971; Jackson
service quality for such options, and to investigate et al., 1984).
the impact of service quality on intentions to use College students were selected as a highly rele-
such options. A secondary research objective was vant population for this study, given that they
to test these models for different situational con- represent a critical consumer segment for fast
ditions. food restaurants (The Wall Street Journal, 1990c).
38 P~A. Dabholkar / lntern. J. o f Research in Marketing 13 (1996) 29-51

Problems of external validity faced by the major- of people or minutes, whichever was easier and
ity of studies using student samples would not be more natural for them to do. A control scenario
relevant for this study, because the results can was used where waiting time was not mentioned.
simply be used by service firms to target college
students. With regard to combining the scenario 3.2. Data collection and sample statistics
approach with college students as respondents,
most students are likely to be familiar with fast The sample consisted of 505 undergraduate
food restaurants and the use of touch screens students from a large university. Almost half the
(e.g., in using ATMs), and therefore should be respondents were men (46.5%) with an average
able to imagine the scenario quite easily. Further- age of 24.80. Women represented 53.5% of the
more, the students were asked to play themselves sample with an average age of 25.31. Less than
in this situation. The scenario method has been one-tenth (8.8 percent) of the respondents re-
found to be particularly successful when subjects ported eating at fast food restaurants 1-2 times a
are required to play themselves instead of acting week, nearly three-fifths (57.6 percent) ate at fast
out unfamiliar roles (e.g., college students or typi- food restaurants 3-5 times a week, and a third
cal consumers asked to imagine they are indus- (33.6 percent) ate 6-8 times a week (apparently,
trial buyers or senior executives) (Eroglu, 1987). some ate at fast food restaurants more than once
Several versions of the scenario were devel- a day).
oped, with different amounts of detail in each Data collection was done in classrooms after
regarding the touch screen option. The realism of informing respondents that the exercise was vol-
each scenario was tested by asking students (not untary and that responses would be anonymous.
the same students as interviewed earlier) to read Questionnaires were handed out so that respon-
a particular version and then rate the realism dents were randomly assigned to treatments. On
using two seven-point Likert scales. The items handing in the questionnaires, respondents were
used were: 'the situation described was realistic' given a one-page sheet (with a matched number
and 'I had no difficulty imagining myself in this to the questionnaire) to conduct checks for treat-
situation' and were based on extensive work done ment manipulation and realism of the scenarios.
on testing the realism of scenarios by Eroglu Of the total respondents, 204 received the low
(1987). The realism checks showed no significant waiting time treatment, 186 received the high
differences between the different versions of the waiting time treatment, and 115 were used as the
scenarios and all were judged to be highly realis- control group whose scenarios had no mention of
tic (rating about 6 on a scale of 1 to 7). It was waiting time.
decided to use the simplest scenario (see Ap-
pendix A) in order to let the respondent imagine 3.3. Measurement
the type of situation (e.g., type of fast food
restaurant, type of touch screen, whether alone 3.3.1. Expected speed of delivery
or with friends, and type of meal order) that was Langeard et al. (1981) measured 'time' in their
most realistic to him/her. study by asking respondents which service option
Different manipulations were pre-tested for (in a given scenario) was faster in their opinion.
waiting time such as the number of people ahead Respondents may have considered 'time' to be
in line or the actual waiting time in minutes. The waiting time for the option a n d / o r speed of
one that worked best in creating a treatment delivery. In addition, given that they used a sin-
effect simply stated, 'You estimate that the wait- gle-item measure, the reliability cannot be ascer-
ing time for using the touch screen to order will tained. Finally, this was a 'relative' item (i.e.,
definitely be longer (shorter) than the waiting measuring time for two options in a comparative
time for placing a verbal order.' Perhaps this sense) and therefore inappropriate for the at-
treatment worked best because it allowed respon- tribute-based model in this study. Hence, a new
dents to imagine waiting time in terms of number scale was developed to measure expected speed
P.A. Dabholkar/ lntern. J. of Research in Marketing 13 (1996) 29-51 39

of delivery. Two seven-point semantic differential not) result in errors in the order,' and '(will/will
items were used following the statement, 'Using not) be reliable.'
the touch screen ordering option will...'. The
endpoints for these items were 'be (fast/slow) 3.3.4. Expected enjoyment
once I'm at the counter' and 'take a (long/short) Previous measures of enjoyment do appear to
time at the register.' The items emphasize 'being be somewhat relevant for this study. Davis et al.
at the (service) counter/register' to indicate that (1992) measured enjoyment using three seven-
the customer has moved up in line and is ready to point semantic differential items based on the
order. This emphasis is aimed at reducing or words 'enjoyable,' 'fun,' and 'pleasant.' In the
eliminating any confounding between waiting time four-item scale used to measure enjoyment in this
and expected speed of delivery. study, the words 'enjoyable' and 'fun' were re-
tained and the descriptors 'entertaining' and 'in-
teresting' were added to capture the novelty as-
3.3.2. Expected ease o f use
pect of this construct, given the context of service
Past measures of ease of use are also not
innovations. The four seven-point semantic dif-
applicable for this study. In their research on the
ferential items were used following the statement,
use of computer technology, Bagozzi (1990) and
'Using the touch screen ordering option will...'.
Davis et al. (1989) measured ease of use in terms
As mentioned, endpoints for these items were
of becoming skillful at using word processing and
' ( b e / n o t be) enjoyable,' ' ( b e / n o t be) fun,'
getting the program to do what the customer
' ( b e / n o t be) entertaining,' and ' ( b e / n o t be) in-
wanted, aspects relevant for continued use of
teresting.'
word processing but not for infrequent use of a
touch screen to order food. Hence, a new scale
3.3.5. Expected control
was developed to measure the ease of using the
Control is viewed as a characteristic directly
touch screen in terms of how easy or effortless it
associated with the use of the technology-based
would be to use this option. The scale consisted
self-service option, independent of situational in-
of four seven-point semantic differential items
fluences. Just as customers can think about an
following the statement, 'Using the touch screen
ordering option and judge its speed of delivery or
ordering option will...'. Endpoints for these
how much they would enjoy the process, so also
items were ' ( b e / n o t be) complicated,' ' ( b e / n o t
they can judge how much control they would have
be) confusing,' 'take (a lot of/little) effort,' and
over the process if they were to use that option.
'require (a lot o f / l i t t l e ) w o r k . '
Some people may feel more in control using the
technology-based self-service option, possibly be-
3.3.3. Expected reliability cause they do not have too high an opinion of the
Expected reliability refers to customer expec- service provider's ability. Other people may feel
tations of how accurately their orders will be less in control using this option, possibly because
filled when they use the technology-based self- they are uncomfortable using technology. A two-
service option. Items from past research measur- item scale was developed to measure expected
ing reliability (Parasuraman et al., 1988) are not control for the specific situation of ordering fast
applicable to this context because those items are food using a touch screen option. One item asked,
linked with promises a n d / o r service personnel. 'How much control will you have over ordering
Hence, a new scale with four seven-point seman- when you use the touch screen ordering option?'.
tic differential items was developed to encompass Responses were indicated on a seven-point se-
this dimension. The items followed the statement, mantic differential scale using end-points 'little
'Using the touch screen ordering o p t i o n . . . ' . The control' and 'a lot of control.' The second item
endpoints for these items were 'means I will used a seven-point Likert scale following the
( g e t / n o t get)just what I ordered,' 'is something I statement, 'the touch screen option will give me
( e x p e c t / d o n ' t expect) to work well,' '(will/will control over ordering a meal.' (Although the word
40 P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51

control appears in both items, this was found to tomer,' 'I like interacting with the person who
be the best wording because words such as 'in provides the service,' 'personal attention by the
charge' or 'dominate' did not correctly convey the service employee is not very important to me,'
meaning of this construct.) and 'it bothers me to use a machine when I could
talk with a person instead.' The responses were
indicated on a seven-point Likert scale.
3.3.6. Prior experience/attitude toward using tech-
nological products
Prior exposure to this particular technology- 3.3.8. Expected service quality
based self-service option was measured directly As Gronroos (1984) suggests, measurement of
by asking respondents if they had ever used a service quality must be defined from the cus-
touch screen to order a meal in a fast food tomer's perspective and must be relevant for the
restaurant. Prior experience with other technolo- situation where it is measured. In this study, it is
gies (e.g., computers, ATMs) was measured using the expected quality of the technology-based
a seven-point frequency scale with endpoints (al- self-service option that is measured and not the
most never/very often). These variables were expected quality of the entire service encounter.
measured to provide descriptive statistics for the Thus, the three items measuring expected service
sample. quality refer specifically to the ordering situation
Existing scales for attitudes towards technol- and not to the whole service encounter where
ogy (e.g., Breakwell et al., 1986) do not measure several factors (such as food, price, and restau-
attitudes towards using technological products. rant conditions) may be relevant. The first item
Hence, a new scale based on attitudinal measure- asked, 'What level of service quality would you
ment (cf. Fishbein and Ajzen, 1975) was devel- receive from the touch screen ordering option?'.
oped for this construct using four seven-point Responses were indicated on a seven-point se-
semantic differential items with the endpoints: mantic differential scale with the endpoints
good/bad, pleasant/unpleasant, harmful/be- '(high/low) quality service.' The second item was
neficial, favorable/unfavorable. These items fol- a seven-point semantic differential scale with
lowed the question, 'How would you describe endpoints '(poor/excellent) service' following the
your feelings toward using technological products statement, 'Using the touch screen ordering op-
in general (e.g., programming a video cassette tion will provide... '. The third item used a
recorder, using a computer)?'. seven-point Likert scale following the statement,
'The touch screen ordering option will provide a
3.3.7. Need for interaction with service employee high level of service quality.'
In attempting to quantify need for interaction,
Langeard et al. (1981) used two items 'human 3.3.9. Intention to use touch screen
contact' and 'dependence on others' which they Intentions were measured using two seven-
viewed as two separate constructs. Their findings point semantic differential items with endpoints
showed that both had on effect on the selection likely/unlikely and possible/impossible (see
of the self-service option. The scale developed in Fishbein and Ajzen, 1975). The stem for these
this study to measure need for interaction com- items asked, 'Would you intend to use the touch
bines and expands on the Langeard et al. items. screen to order fast food?'
The focus is on 'interacting with a service em-
ployee'; the phrases 'human contact' and 'depen-
dence on others' may have been too brief to 3.3.10. Manipulation and realism checks
precisely measure this need within service en- The manipulation check for waiting time asked
counters. Four items were developed to measure respondents to think back to the scenario and
this construct: 'human contact in providing ser- recall whether 'the waiting time for the touch
vices makes the process enjoyable for the cus- screen was longer than that for ordering verbally.'
P.A. Dabholkar / lntern. J. of Research in Marketing 13 (1996) 29-51 41

Responses were indicated on a seven-point Likert 4. Results


scale. Realism checks were measured on the same
sheet using two seven-point Likert items. The 4.1. Manipulation and other checks
statements in these items were: 'the situation
described was realistic' and 'I had no difficulty Manipulation checks were made for waiting
imagining myself in the situation.' time using ANOVA and found that the treatment

Table 2
Correlations among variables for different waiting time conditions
Int Qual Speed Ease Rely Enjoy Cont Att Need
Low waiting time (n = 204)
Mean 5.71 4.45 5,16 4.86 5.36 5.21 5.35 5.90 4,57
Std. Dev. 1.30 0.90 1.35 1.37 1.19 1.12 1.33 0.82 1.08
Int 1.00
Qual 0.55 1.00
Speed 0.38 0.20 1,00
Ease 0.37 0.22 0.62 1.00
Rely 0.50 0.36 0.56 0.55 1.00
Enjoy 0.46 0.45 0.22 0.16 0.40 1.00
Cont 0.52 0.34 0.37 0.40 0.59 0.33 1.00
Att 0.40 0.22 0.20 0.34 0.32 0.25 0,37 1.00
Need -0.31 -0.31 -0.01 -0.18 -0.25 -0.14 -0,29 -0.27 1.00

High waiting time (n = 186)


Mean 4.77 4.24 4.24 4.40 5.30 4.89 5.36 5.81 4.56
Std. Dev. 1.79 1.07 1.58 1.39 1.09 1.33 1,33 0.87 1.15
Int 1.00
Qual 0.63 1.00
Speed 0.26 0.38 1.00
Ease 0.32 0.37 0.54 1.00
Rely 0.18 0.21 0.30 0.34 1.00
Enjoy 0.48 0.46 0.17 0.10 0.28 1.00
Cont 0.27 0.30 0.28 0.18 0.58 0.29 1.00
Att 0.11 0.22 0.04 0.05 0.13 0.24 0.24 1,00
Need -0.32 -0.31 -0.23 -0.29 -0.23 -0.19 -0.27 -0.04 1.00

Control group (n = 115)


Mean 5.75 4.63 4.89 4.65 5.42 5.15 5.51 5.96 4.38
Std. Dev. 1.45 1.02 1.48 1.40 1.15 1.41 1,22 0.85 1.31
Int 1.00
Qual 0.57 1.00
Speed 0.33 0.31 1.00
Ease 0.45 0.43 0.61 1.00
Rely 0.50 0.37 0.49 0.61 1.00
Enjoy 0.45 0.35 0.12 0.23 0.28 1.00
Cont 0.62 0.40 0.39 0.37 0.52 0.17 1.00
Att 0.39 0,32 0.20 0.18 0.24 0.36 0.37 1.00
Need -0.54 -0.47 -0.21 -0.30 -0.35 -0.23 -0.51 -0.35 1.00

Int: intention to use technology-based self-service option. Qual: service quality of technology-based self-service option. Speed:
speed of delivery. Ease: ease of use. Rely: reliability. Enjoy: enjoyment. Cont: control. Att: attitude toward using technological
products. Need: need for interaction with service employee.
42 P.4. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51

worked well ( F = 133.07, p < 0.001). The means Table 3


of the manipulation check items were 2.74 for low A test of alternative models of service quality for different
waiting time conditions
waiting time, 5.32 for high waiting time, and 3.80
for the control group (on a scale of 1 to 7). Attribute- Overall Combined
based affect model
Scheffe's paired comparison tests showed that all model model
three means differed significantly at p < 0.05.
To ensure that there was no confounding be- Low waiting time (n = 204)
X2 187.26 62.54 343.49
tween waiting time and speed of delivery, t-tests df 102 39 222
were conducted for speed of delivery under high AGFI 0.86 0.91 0.83
and low waiting time. The means for speed of R 2 (service quality) 0.55 0.23 0.60
delivery were 5.17 (low waiting time) and 4.44 R 2 (intention) 0.65 0.62 0.65
(high waiting time), but this difference was not High waiting time (n ~ 186)
statistically significant ( F = 1.24, n.s.). Thus, there gz 125.43 49.91 277.53
was no confounding between waiting time and df 102 39 222
expected speed of delivery. AGFI 0.89 0.92 0.86
Means for the realism checks (using two items) R z (service quality) 0.73 0.17 0.73
R 2 (intention) 0.67 0.66 0.67
rated 5.94 on a scale of 1 to 7, showing that
respondents found all three scenarios to be very Control group (n = 115)
realistic. X2 141.15 51.02 290.39
df 102 39 222
AGFI 0.83 0.88 0.79
4.2. Descriptive statistics R 2 (service quality) 0.70 0.37 0.7,3
R 2 (intention) 0.50 0.50 0.52
About 68.9 percent of the respondents said
they would use a touch screen to order fast food.
Yet, prior exposure to touch screens in fast food
restaurants was almost non-existent, which is un- ing support for the constructs conceptualized in
derstandable given that this is a new form of the study. Construct reliabilities computed for
ordering option. Only eighteen people out of 505 five of these scales from the results of separate
(3.6 percent) had used this ordering option. The confirmatory factor analyses were 0.91 (Ease of
level of prior exposure to using technological Use), 0.88 (Reliability), 0.91 (Enjoyment), 0.86
products in general was neither high nor low with (Attitude toward Using Technological Products),
a m e a n rating of 4.16 (on a scale of 1 to 7), but and 0.83 (Need for Interaction with Service Em-
prior exposure to using automated teller ma- ployee). The remaining four scales had fewer
chines in particular, was quite high at 5.45. Means, than four items; hence, Cronbach's alphas were
standard deviations, and correlations for all the computed for these scales and were 0.81 (Speed
constructs are presented separately for the three of Delivery), 0.69 (Control), 0.85 (Service Quality),
situational conditions (see Table 2). and 0.90 (Intention). Thus, reliabilities ranged
from 0.69 to 0.91, indicating that all the scales
4.3. M e a s u r e validity developed in this study were acceptable and reli-
able measures (Nunnally, 1978).
Confirmatory factor analysis using L I S R E L 7
(Joreskog and Sorbom, 1989) was conducted for 4.4. A c o m p a r i s o n o f m o d e l s
all the scales together. Confirmatory factor analy-
sis is preferred over exploratory factor analysis It was decided to use structural equations
because it is theory-based, accounts for measure- modeling using L I S R E L 7 to compare the at-
m e n t error, and tests for unidimensionality. The tribute-based and overall affect models. This
results showed a satisfactory fit to the data, with method of analysis has several advantages over
X 2 = 322.42, df = 255, and A G F I = 0.93, indicat- typical multivariate analysis. Whereas two regres-
P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51 43

Table 4
A test of the attribute-based model
P a r a m e t e r estimates and significance for individual hypotheses
HI H2 H3 H4 H5 H8
Low waiting time (n = 204) 0.07 (n.s.) 0.07 (n.s.) - 0 . 0 2 (n.s.) 0.48 ( p < 0.001) 0.32 ( p < 0.05) 1.31 ( p < 0.001)
High waiting time (n = 186) 0.06 (n.s.) 0.35 ( p < 0.01) - 0 . 5 7 (n.s.) 0.52 ( p < 0.001) 0.62 ( p < 0.05) 1.24 ( p < 0.001)
Control group (n = l l 5 ) -0.13(n.s.) 0.25(p<0.05) -0.11(n.s.) 0.47(p<0.001) 0.69(p<0.001) 1.55(p<0,001)

HI: speed of delivery ~ service quality of TBSSO. H2: ease of use ~ service quality of TBSSO. H3: reliability ~ service quality of
TBSSO. H4: enjoyment ~ service quality of TBSSO. H5: control ~ service quality of TBSSO. H8: service quality of TBSSO
intention to use TBSSO.
TBSSO: technology-based self-service option.
The p-values are based on t-values in the L I S R E L output.

sion equations are needed to test each model in plained far more of the variance for service qual-
this paper, structural equations modeling allows ity than did the overall affect model. The R 2 for
all variables to be included in one model and service quality in the attribute-based model as
tests for model fit as well as individual hypothe- compared to the R 2 for service quality in the
ses. It accounts for measurement error thus re- overall affect model was 0.55 vs. 0.23 for low
vealing relationships that may be otherwise ob- waiting time, 0.73 vs. 0.17 for high waiting time,
scured. Finally, an examination of modification and 0.70 vs. 0.37 for the control group (see Table
indices can reveal other relationships not hypoth- 3). Combining both models was even more reveal-
esized in the model. ing in terms of variance explained. The combined
Model testing was done separately for each model using all the attributes and dispositions as
situational group: low waiting time, high waiting exogenous variables showed that the increase in
time, and the control group. Both models had R 2 on including the overall affect variables was
good fits with the data irrespective of the situa- either non-existent or merely incremental over
tional condition (see Table 3). Although the AGFI the variance explained by the attribute-based
values were slightly higher for the overall affect model alone. The R 2 for the combined model
model as compared with the attribute-based was 0.60 for low waiting time, 0.73 for high wait-
model (0.91 vs. 0.86 for low waiting time, 0.92 vs. ing time, and 0.73 for the control group.
0.89 for high waiting time, and 0.88 vs. 0.83 for Examining the parameter estimates and t-val-
the control group), this difference may be due to ues for the individual hypotheses in the
the fact that the overall affect model had fewer attribute-based model, it was found that hypothe-
than half the number of indicators as the at- ses H4, H5, and H8 were supported for all three
tribute-based model. 4 As the A G F I values for situational conditions (see Table 4). Moreover,
both models under all situational conditions are the size of the parameter estimates suggest a
within an acceptable range, the conclusion is that strong, positive effect of enjoyment and control
both models fit well. on service quality and a very strong, positive
An interesting finding is that the attribute- effect of service quality on intentions. Hypothesis
based model in each of the three situation ex- H2 was supported for the high waiting time and
control groups but not for the low waiting time
situation. The size of the parameter estimates
4 Notice for example that the A G F I ' s for the combined suggests that ease of use had a strong, positive
model u n d e r all three situational conditions are lower than effect on service quality for most people (except
for either model alone (see Table 3). Given that the same
constructs are used in the combined model as in the other two
those who faced a short waiting time). Hypothe-
models, the reduction in A G F I is likely to be due to the ses H1 and H3 were not supported for any group
increase in the n u m b e r of indicators in the model. suggesting that speed of delivery and reliability
44 PM. Dabholkar ~Intern. J. of Research in Marketing 13 (1996) 29-51

Table 5
A test of the overall affect model
Parameter estimates and significance for individual hypotheses
H6 H7 H8
Low waiting time (n = 204) 0.64 ( p < 0.001) -0.14 (p < 0.05) 1.38 (p < 0.001)
High waiting time (n = 186) 0.47 ( p < 0.01) -0.34 (p < 0.001) 1.05 ( p < 0.001)
Control group (n = 115) 0.66 (p < 0.001) -0.30 ( p < 0.001) 1.52 ( p < 0.001)

H6: attitude toward using technological products --* service quality of TBSSO. H7: need for interaction with service employee
service quality of TBSSO. H8: service quality of TBSSO --* intention to use TBSSO.
TBSSO: technology-based self-service option.
The p-values are based on t-values in the LISREL output.

did not have an effect on service quality under both hypotheses H10a and H10b were supported.
any situation. High modification indices sug- Parameter estimates suggested that waiting time
gested direct effects on intentions from enjoy- had a very strong, negative effect on intention
ment and control (in addition to their indirect (b = - 1.07, p < 0.001) and a fairly strong, nega-
effects through service quality) for the low wait- tive effect on expected service quality (b = - 0.32,
ing time and control groups. High modification p <0.01). However, R 2 values were not ex-
indices also suggested direct effects on intentions tremely high especially for service quality. The R 2
from speed of delivery, control, and reliability for was 0.02 for service quality and 0.10 for inten-
the high waiting time group. sions. 5
Examining the parameter estimates and t-val-
ues for the individual hypotheses in the overall
affect model, it was found that all the hypotheses 5. Discussion
H6, H7, and H8 were supported for all three
situational conditions (see Table 5). Thus, atti- 5.1. Theoretical contribution
tude toward using technological products had a
positive effect on service quality, need for inter- Although technology-based self-service options
action had a negative effect on service quality, are becoming more feasible and have obvious
and service quality had a positive effect on inten- monetary benefits for service firms, consumer
tions as predicted, for all situations tested. Unlike evaluations of service quality for such options
the attribute-based model, no direct effects on have not been investigated. This research pro-
intentions were suggested by modification in- poses two alternative models of service quality -
dices. an attribute-based model and an overall affect
So far all the analyses had been conducted model - and represents one of the first attempts
separately for each waiting time condition. How- to apply consumer decision making concepts to
ever, to test the effect of waiting time on inten- the services area. Although tested here for expec-
tion to use and expected service quality of the tations of service quality before a service en-
touch screen option, the entire sample (minus the counter, the models are equally applicable for
control group) had to be used. Waiting time was
modeled as an exogenous (independent) variable
and service quality and intentions as the endoge-
nous (dependent) variables. The model fit was 5 A second model was run which included the effect from
service quality to intentions. This model being 'equivalent' to
excellent ( g 2 = 8.01, df = 4, and A G F I = 0.97).
the first model (in terms of structural equations) had the same
Examining the parameter estimates and t-values overall fit. However, the R 2 for intentions increased to 0.69
for the individual hypotheses, it was found that with the inclusion of service quality as a causal variable.
P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51 45

perceptions of service quality after a service en- also directly increases intention to use the option.
counter. It would appear that faced with a long waiting
Consumers appear to favor the attribute-based time, the impact of the potential enjoyment or
model in forming evaluations of service quality fun from this option is somewhat less relevant.
for technology-based self-service options, based A third attribute, ease of use, was also found
on the finding that cognitive evaluations of ser- to be an important determinant of service quality
vice delivery attributes explain a sizeable portion but only for the high waiting time and control
of the variance in expected service quality. Al- groups. This suggests that facing a shorter waiting
though the overall affect model is also supported, time for this option, ease of use would not be
it does not add further explanatory power to the critical in evaluating service quality. The last two
attribute-based model. These results support con- attributes, speed of delivery and reliability, did
sumer decision making theory wherein unfamil- not influence evaluations of service quality under
iarity with a situation encourages cognitive evalu- any situational condition. However, they did have
ation (Bettman and Park, 1980; Park and Lessig, direct effects on intentions for the high waiting
1981) as does the availability of a small number of time group.
options (Lussier and Olshavsky, 1979). It is possible that speed of delivery and relia-
The study tests these models under different bility had no effect on service quality in the study
situational conditions to investigate the impact of because of the intercorrelations among indepen-
situational factors on service quality and cus- dent variables. Speed of delivery was highly corre-
tomer participation in service delivery. The pref- lated with ease of use and reliability was highly
erence for the cognitive over the affective model correlated with control for all three situational
is consistent over all situational conditions, but is groups. These intercorrelations are not unex-
m o s t pronounced for the high waiting time group pected because the constructs for speed of deliv-
(given that there is n o increase in variance ex- ery and ease of use are closely related as are the
plained on adding the affect variables to the constructs for reliability and control. If a service
attribute-based model). An explanation may be delivery option is easy to use, it is also likely that
that facing a long waiting time, people are even it will be fast. If a service delivery option is seen
more apt to evaluate the option attribute by as reliable, then the customer will be likely to
attribute to decide if it is worth waiting for. expect greater control using this option. Unfortu-
The individual hypotheses in the models tested nately, as in typical multivariate analysis, LIS-
in this study are based on past research on ser- R E L is also sensitive to multicollinearity and
vice delivery, self-service, and the use of techno- hence it is possible that the effects of speed of
logical products as well as on qualitative research delivery and reliability do not emerge (given that
specifically conducted for this study. The study the effects of ease of use and control do).
found that enjoyment and control were important The direct effect of speed of delivery on inten-
determinants of service quality under all three tions for the high waiting time group is certainly
situational conditions. Moreover, the results sug- understandable. Facing a long waiting time, a
gested direct effects on intentions from control customer's decision whether or not to use the
under all three situational conditions and from option will be influenced by how fast s / h e thinks
enjoyment for the low waiting time and control the option will be (once s / h e reaches the regis-
groups. Thus, feeling in control over the process ter). The reason for the direct effect of reliability
of service delivery, enhances consumer evalua- on intentions is not as clear although a similar
tions of this process and also directly impacts rationale (as for speed of delivery) is possible.
intentions to use the option. The potential enjoy- Although the overall affect model did not add
ment from using this service delivery option en- to the variance explained by the attribute-based
hances customer evaluations of this option and model, the results did build on previous research.
faced with a shorter or similar waiting time (com- The study found that attitude toward using tech-
pared to other options), this potential enjoyment nological products had a positive effect on evalu-
46 P.A. Dabholkar / Intern. Z of Research in Marketing 13 (1996) 29-51

ations of technology-based self-service options, lection of a service delivery option, evaluations of


supporting and extending previous research on service quality may be formed without regard to
categorized affect (Dickerson and Gentry, 1983; situational factors because customers would ex-
Ledingham, 1984) and that need for interaction pect such factors to differ across incidents of
with a service employee was an important deter- service delivery.
minant of service quality which past studies (Lan-
geard et al., 1981) had predicted but not found.
The finding that evaluations of service quality 5.2. Strategic implications for service firms
have a strong positive influence on intended se-
lection of ordering option justifies the attention The finding that more than two-thirds of the
given to service quality issues in the literature. It student respondents said they would select touch
verifies the intuitive connection that if people screen ordering over verbal ordering should be
expect a service delivery option to be of high quite encouraging to both fast food restaurant
quality, they will use it. At the same time, it is owners and touch screen manufacturers. The re-
curious that while roughly two-thirds of the vari- suits provide strategic direction by suggesting that
ance in intention is explained (by either model) touch screens installed in restaurants located near
under the low and high waiting time conditions college campuses would be well received. Given
(see R 2 in Table 3), only half the variance in how frequently college students eat at fast food
intention is explained (by either model) for the restaurants, the investment would be quickly re-
control group. A possible explanation for these covered. Further, as students represent a large
findings is that for the low and high waiting time proportion of fast food consumers (The Wall
conditions, an important situational factor is Street Journal, 1990c), it is possible that these
known and this is taken into consideration in results may be closely replicated in a sample
forming intentions to use the option. For the drawn from the general population. Similar re-
control group, no situational factor is mentioned sults may be obtained using other samples whose
and respondents may reason that irrespective of members are familiar with technology and sensi-
their evaluation of service quality, their actual tive to waiting time, such as white collar workers
intentions would at least partially be influenced and professionals.
by the situational condition at the time (which is Given that the attribute-based model explains
unknown at present). Thus, for this group, service most of the variance in service quality suggests
quality would account for less of the variance in that service firms can design and promote at-
intention than in groups where situational factors tributes of the service delivery that will lead to
are known and taken into consideration. better evaluations of service quality. Had the
It was found that waiting time as a situational overall affect model accounted for most of the
factor influenced intentions to use a technology- variance in service quality, this would have sug-
based self-service option as expected, supporting gested that service firms would need an immedi-
previous research (Maister, 1985; Clemmer and ate focus on changing consumer attitudes toward
Schneider, 1989). Although waiting time also had using technology and toward service providers in
an effect on evaluations of service quality for order to improve evaluations of service quality.
technology-based service delivery, this effect was Most firms would view this as a formidable task.
of practical insignificance. One explanation for Thus, the results of the comparison of models are
these findings may be that waiting time may only encouraging for service firms in terms of strategic
influence evaluations of service quality if a nega- direction.
tive emotional reaction has occurred while wait- The finding that enjoyment and control
ing (see Taylor, 1994) and an emotional reaction strongly influence service quality as well as inten-
with waiting is less likely with a scenario-based tions clearly suggests that these two attributes
study. Another explanation may be that although should be emphasized in service design and pro-
situational influences impact the customer's se- motion. The fun or novelty aspect needs to be
P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51 47

built into the service design, perhaps through the toward using technological products by encourag-
use of colorful icons and humorous directions. In ing greater use of technology in general. Need for
terms of promotion, this aspect could be empha- interaction was found to be an important nega-
sized by contrasting people having 'fun' using tive d e t e r m i n a n t of service quality for
such options with people looking quite bored technology-based self-service options. When ad-
when interacting with service employees. Simi- vertising to audiences who have a low need for
larly, given the importance of control to respon- interaction with service employees, firms could
dents, service design must ensure that customers emphasize that customers can get what they want
are not overpowered by the technology. In addi- without having to interact with anyone. However,
tion, ads can show the customer 'in control' when when advertising to audiences with a high need
using the technology-based self-service option, by for interaction with service employees, firms
having them express such thoughts or suggest it would do better to stress the availability of em-
through facial expression. ployees if needed even when using the technol-
Ease of use was also a determinant of service ogy-based self-service options.
quality for the high waiting time and control The negative impact of waiting time on inten-
groups. Hence, service firms should design user- tions to use technology-based self-service options
friendly and easy to use technology-based self- is certainly not surprising, but it does have impli-
service options that the consumer will evaluate cations for service design and promotion. A suffi-
more highly. Promoting user-friendliness can only cient number of automated, self-service kiosks
help further enhance customer evaluations of should be installed based on demand, to keep
these options. waiting time to a minimum. Special promotions
The other two attributes, speed of service de- may be offered at off-peak times to reduce wait-
livery and reliability, had no effect on service ing time during peak hours. The lack of waiting
quality but did have a direct effect on intensions time for technology-based service delivery options
for the high waiting time group. As discussed, the at certain service sites may be promoted to woo
effect of these two variables on service quality customers to those locations.
may not have merged due to multicollinearity
(specifically between speed of delivery and ease
of use and between reliability and control). In any
5.3. Limitations and f u t u r e research
case, given that they do impact the decisions of
people facing long waiting lines, these attributes
also need to be incorporated into design and While the scenario and questionnaire ap-
promotion. Service firms should use a technology proach was appropriate for this study for reasons
that is reliable and a process that is quick and explained, expectations of service quality or in-
promote the same. For example, National Car tentions to use technology-based self-service op-
Rental's ads for their automated, car rental ma- tions as indicated in surveys may not translate
chines located in airports already emphasize into actual behavior. Future research could inves-
speed of service delivery and reliability. tigate the link between expected service quality
Although the overall affect model did not add and selection of technology-based self-service op-
much to the attribute-based model in terms of tions in an actual service encounter. There is no
predictive power, the findings related to this doubt that actually seeing how service delivery
model have long-range implications for service options work would evoke more reliable re-
firms. Generalized attitudes toward using techno- sponses from consumers. However, as most re-
logical products had a positive effect on evalua- search involves a tradeoff, researchers conducting
tions of service quality for technology-based self- a field study would have to acknowledge the
service options as predicted. Hence, service firms limitations imposed by a shorter questionnaire,
that plan to install such options would do well anticipate nonrespondent bias, and accept that
over the long term to promote favorable attitudes situational factors (such as waiting lines for dif-
48 P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51

ferent options) may be somewhat beyond the and ease of use on evaluations of the service
researcher's control. delivery. On the other hand, an established tech-
Although highly relevant for the context, a nology may not be evaluated quite as highly on
student sample does not offer much variance in the enjoyment factor, a large component of which
age, education, and attitude toward using tech- may be its novelty aspect.
nology. Future researchers could use a broader Finally, certain attributes that may be relevant
based sample to investigate the effect of demo- for other contexts did not emerge from the quali-
graphic differences, as well as differences in atti- tative research and were therefore excluded. The
tudes toward using technology, on expectations of context selected for this study could be viewed as
service quality. Age and cultural differences in a a low-risk situation and hence risk was not specif-
more representative sample may also be reflected ically modeled into the framework. Yet, to some
in different levels of the need for interaction with extent, the model encompasses performance risk
service employees. In addition to testing for dif- within reliability and social risk within ease of
ferent samples, the models developed in this study use. Similarly, given that the context did not
may be tested in other service contexts where require a high level of expertise to perform the
on-site, technology-based self-service options service, expertise was not specifically modeled
could be installed. Examples of such contexts into the framework. Again, to an extent, expertise
include post offices with stamp machines, airports could be viewed as encompassed in estimates of
with ticketing and car rental machines, hotels the expected ease of use. For other contexts,
with in-room closed-circuit check-out, and de- however, such as monitoring one's health or buy-
partment stores with electronic, self-service op- ing insurance through automation, consumers may
tions. expect that using an automated system would be
The only situational influence addressed in quite risky and would require significant expertise
this study was waiting time. Other situational on their part (Insurance Sales, 1988); conse-
influences may be investigated in future studies. quently, for such contexts, risk and expertise may
Examples might be time pressures (such as being have to be modeled as separate dimensions. Also,
in a hurry) or the presence of other customers in this study, physical environment and price were
(such as crowding). Crowding may be more likely considered to be the same for the technology-
to have an impact on the general population than based self-service as well as the service delivered
on students who are somewhat used to crowding. by the employee (based on test marketing plans
It may also work with more complex technologies by a fast food chain). Hence these factors were
where crowded conditions are more likely to cause not modeled into the framework. However, for
stress. some contexts (e.g., an ATM at a location outside
Future studies can continue to apply consumer the bank), physical environment could be impor-
decision-making theory to services marketing as tant in terms of convenience and safety and for
was done in this study by testing alternative mod- other contexts (e.g., paying a fee for using an
els of service quality for a variety of contexts. It is ATM), price could be relevant. Future studies
possible that the overall affect model would fit could investigate factors such as physical environ-
better for certain types of services where cus- ment and price for contexts where they are ex-
tomers are less able to evaluate alternatives or pected to be different across service delivery op-
where strong emotions are aroused (e.g., health tions.
care). For other contexts where the attribute-
based model explains most of the variance in
service quality as in this study, the attributes
included in this framework may impact evalua- 6. For further reading
tions of service quality differently. For example, a
technology that is completely new to the cus- For further reading, see World of Banking,
tomer may increase the importance of reliability 1990.
P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51 49

Appendix A. Stimulus for questionnaire in service organizations, Human Resource Management 25


(3), 371-383.
BreakweU, G.M., C. Fife-Schaw, T. Lee and J. Spencer, 1986,
A. 1. Situation
Attitudes towards new technology in relation to social
beliefs and group memberships: A preliminary investiga-
You are in a fast food restaurant for lunch. tion, Current Psychological Research and Reviews 5 (1),
Approaching the counter, you find that you have 34-47.
two ordering options: you may order verbally as Business Week, 1990, Service pay is fueling inflation 26 (April
usual or use the newly installed touch screen 23).
machine. (The touch screen is located on the Buzzell, R.D. and B.T. Gale, 1987, The PIMS principles:
Linking strategy to performance (The Free Press, New
counter and has directions for use and the com- York).
plete m e n u on the screen itself. Ordering is done Chain Store Age Executive, 1987, Vol. 63 (12), Technology:
by touching the appropriate boxes on the screen At home on the selling floor, 35-36.
and items can be re-entered if you make a mis- Chase, R.B., 1978, Where does the customer fit in a service
take or change your mind.) Both options have the operation? Harvard Business Review 56 (November-De-
cember), 137-142.
same m e n u at the same prices and allow you to Chebat, J. and P. Filiatrault, 1993, The impact of waiting in
personalize your order (e.g., 'hold the mustard,' line on consumers, International Journal of Bank Market-
etc.). In each case you pay the cashier after ing 11 (2), 35-40.
placing the order and the cashier hands you the Clemmer, E.C. and B. Schneider, 1989, Towards understand-
food you ordered when it is ready. You estimate ing and controlling dissatisfaction with waiting, Report
89-115 (Marketing Science Institute, Cambridge, MA).
that the waiting time for using the touch screen Cowles, D., 1989, Consumer perceptions of interactive media,
to order will definitely be longer (shorter) than Journal of Broadcasting and Electronic Media 33 (1),
the waiting time for placing a verbal order. 83-89.
Note: The last sentence in the scenario is used Cowles, D. and L.A. Crosby, 1990, Consumer acceptance of
to manipulate the treatment, waiting time, and is interactive media, The Services Industries Journal 10 (3),
521-540.
omitted for the control group.
Cronin, J.J. and S.A. Taylor, 1992, Measuring service quality:
A reexamination and extension, Journal of Marketing 56
(3), 55-68.
References Czepiel, J.A., 1980, Managing customer satisfaction in con-
sumer service businesses, Report No. 80-107 (Marketing
Bagozzi, R.P., 1990, Buyer behavior models for technological Science Institute, Cambridge, MA).
products and services: A critique and proposal, in: W. Dabholkar, P.A., 1994a, Technology-based service delivery: A
Johnston, ed., Advances in telecommunications manage- classification scheme for developing marketing strategies,
ment, Vol. 2 (JAI Press, CT), 43-69. in: T.A. Swartz, D.E. Bowen and S.W. Brown, eds., Ad-
Bateson, J.E.G., 1985, Self-service consumer: An exploratory vances in services marketing and management, Vol. 3 (JAI
study, Journal of Retailing 61 (3), 49-76. Press, Greenwich, CT), 239-269.
Bateson, J.E.G. and M.K.M. Hui, 1987, Perceived control as a Dabholkar, P.A., 1994b, Incorporating choice into an attitudi-
crucial perceptual dimension of the service experience: An nal framework: Analyzing models of mental comparison
experimental study, in: C.F. Surprenant, ed., Add value to processes, Journal of Consumer Research 21 (June), 100-
your service (American Marketing Association, Chicago), 118.
187-192. Dabholkar, P.A., 1990, How to improve perceived service
Berry, L.L., D.R. Bennett and C.W. Brown, 1989, Service quality by increasing customer participation, in: B.J. Dun-
quality: A profit strategy for financial institutions (Dow lap, ed., Developments in marketing science, XIII
Jones-Irwin, Homewood, IL). (Academy of Marketing Science, Cullowhee, NC), 483-487.
Bettman, J.R. and C.W. Park, 1980, Effects of prior knowl- Davis, F.D., R.P. Bagozzi and P.R. Warshaw, 1989, User
edge and experience and phase of the choice process on acceptance of computer technology: A comparison of two
consumer decision processes: A protocol analysis, Journal theoretical models, Management Science 35 (8), 982-1003.
of Consumer Research 7 (December), 234-248. Davis, F.D., R.P. Bagozzi and P.R. Warshaw, 1992, Extrinsic
Bitner, M.J., B.H. Booms and M.S. Tetreault, 1990, The and intrinsic motivation to use computers in the work-
service encounter: Diagnosing favorable and unfavorable place, Journal of Applied Social Psychology 22 (14), 1109-
incidents, Journal of Marketing 54 (January), 71-84. 1130.
Bowen, D.E., 1986, Managing customers as human resources Dickerson, M.D. and J.W. Gentry, 1983, Characteristics of
50 P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51

adopters and non-adopters of home computers, Journal of note on the perception of time in consumer behavior,
Consumer Research 10 (September), 225-235. Journal of Consumer Research 11 (June), 615-618.
Dube-Rioux, L., B.H. Schmitt and F. Leclerc, 1989, Consumer Insurance Sales, 1988, Vol. 131 (8), High-tech and no-touch:
reactions to waiting: When delays affect the perception of Is person-to-person selling obsolete?, 26-28.
service quality, in: T. Srull, ed., Advances in consumer Jacoby, J. and J.C. Olson, 1985, Perceived quality (Lexington
research, Vol. 16 (Association for Consumer Research, Books, Lexington, MA), xi-xiv.
Provo, UT), 59-63. Jackson, D.W., J. Keith and R.K. Burdick, 1984, Purchasing
Eroglu, S.A., 1987, The scenario method: A theoretical, not agents' perceptions: A situational approach, Journal of
theatrical, approach, in: S.P. Douglas et al., eds., Proceed- Marketing 48 (4), 75-83.
ings of the AMA Summer Educators' Conference, 236 Johnson, M.D., 1984, Consumer choice strategies for compar-
(American Marketing Association, Chicago). ing noncomparable alternatives, Journal of Consumer Re-
Evans, K.R. and S.W. Brown, 1988, Strategic options for search 11 (December), 741-753.
service delivery systems, in: C.A. Ingene and G.L. Frazier, Joreskog, K.G. and D. Sorbom, 1989, LISREL7: A guide to
eds., Proceedings of the AMA Summer Educators' Con- the program and applications, 2nd ed. (SPSS Inc., Chicago,
ference (American Marketing Association, Chicago), 207- IL).
212. Kelley, S.W., J.H. Donnelly, Jr. and S.J. Skinner, 1990, Cus-
Fishbein, M. and I. Ajzen, 1975, Belief, attitude, intention, tomer participation in service production and delivery,
behavior: An introduction to theory and research (Ad- Journal of Retailing 66 (3), 315-335.
dison-Wesley Publishing Co., Reading, MA). Kolesar, P., 1984, Stalking the endangered CAT: A queuing
Fiske, S.T., 1982, Schema-triggered affect, in: A.H. Hastorf analysis of congestion at automatic teller machines, Inter-
and A.H. Isen, eds., Cognitive social psychology (Elsevier, faces 14 (6), 16-26.
New York), 55-78. Langeard, E., J.E.G. Bateson, C.H. Lovelock and P. Eiglier,
Forman, A.M. and V. Sriram, 1991, The depersonalization of 1981, Marketing of services: New insights from consumers
retailing: Its impact on the 'lonely' consumer, Journal of and managers, Report No. 81-104 (Marketing Science
Retailing 67 (2), 226-243. Institute, Cambridge, MA).
Glass, D.C. and J.E. Singer, 1972, Urban stress (Academic Langer, E.J., 1975, The illusion of control, Journal of Person-
Press, New York). ality and Social Psychology 32 (2), 311-328.
Gronroos, C., 1982, An applied service marketing theory, Langer, E.J. and S. Saegert, 1977, Crowding and cognitive
European Journal of Marketing 16 (7), 30-41. control, Journal of Personality and Social Psychology 35
Gronroos, C., 1983, Innovative marketing strategies and orga- (3), 175-182.
nization structures for service firms, in: L.L. Berry, G.L. Ledingham, J.A., 1984, Are consumers ready for the informa-
Shostack and G.D. Upah, eds., Emerging perspectives on tion age?, Journal of Advertising Research 24 (4), 31-37.
services marketing (American Marketing Association, Lovelock, C.H. and R.F. Young, 1979, Look to consumers to
Chicago), 9-21. increase productivity, Harvard Business Review 57 (May-
Gronroos, C., 1984, A service quality model and its marketing June) 168-178.
implications, European Journal of Marketing 18 (4), 36-44. Lussier, D. and R. Olshavsky, 1979, Task complexity and
Gronroos, C., 1993, From scientific management to service contingent processing in brand choice, Journal of Con-
management, International Journal of Service Industry sumer Research 6 (September), 154-165.
Management 5 (1), 5-20. Maister, D.H., 1985, The psychology of waiting lines, in: J.A.
Guiry, M., 1992, Consumer and employee roles in service Czepiel et al., eds., The service encounter: Managing
encounters, in: J.F. Sherry, Jr. and B. Sternthal, eds., employee/customer interaction in service businesses
Advances in consumer research, Vol. 19 (Association for (Lexington Books, Lexington, MA), 113-126.
Consumer Research, Provo, UT) 666-672. Marketing News, 1990, November 12, Arby's adds technology
Gummesson, E., 1993, Quality management in service organi- to boost service, 1.
zations (International Service Quality Association, New Mills, P.K. and D.J. Moberg, 1982, Perspectives on the tech-
York). nology of service operations, Academy of Management
Holbrook, M.B. and E.C. Hirschman, 1982, The experiential Review 7 (3), 467-478.
aspects of consumption: Consumer fantasies, feelings and Mills, P.K. and J.H. Morris, 1986, Clients as 'partial' employ-
fun, Journal of Consumer Research 9 (September) 132- ees of service organizations: Role development in client
140. participation, Academy of Management Review 11 (4),
Holbrook, M.B., R.W. Chestnut, T.A. Oliva and E.A. Green- 726-735.
leaf, 1984, Play as a consumption experience: The roles of Mixon, D., 1971, Behavior analysis teaching subjects as actors
emotions, performance and personality in the enjoyment rather than organizations, Journal of the Theory for Social
of games, Journal of Consumer Research 11 (September), Behavior 1, 19-31.
728-739. Nation's Restaurant News, 1991, March 11, Fast feeders turn
Hornik, J., 1984, Subjective and objective time measures: A on video menu systems, 7.
P.A. Dabholkar / Intern. J. of Research in Marketing 13 (1996) 29-51 51

Nunnally, J.C., 1978, Introduction to psychological measure- The New York Times, 1989, April 9, The powerful push for
ment (McGraw-Hill, New York). self-service, $3-1.
Parasuraman, A., V.A. Zeithaml and L.L. Berry, 1988, The Scarborough Market Report, 1986, (The Journal and
SERVQUAL: A multiple-item scale for measuring con- Constitution, Atlanta).
sumer perceptions of service quality, Journal of Retailing The Wall Street Journal, 1990a, November 12, Many con-
64 (1), 12-40. sumers expect better service, B1.
Park, C.W. and V.P. Lessig, 1981, Familiarity and its impact The Wall Street Journal, 1990b, December 7, Their business
on consumer decision biases and heuristics, Journal of is on the line, B1.
Consumer Research 8 (September), 223-230. The Wall Street Journal, 1990c, November 27, Fast-food
Schneider, B. and D.E. Bowen, 1985, Employee and customer chains invade college eateries, B1.
perceptions of service in banks, Journal of Applied Psy- Van Gorder, B.E., 1990, Satisfying the customer of the '90s,
chology 70 (3), 423-433. Credit 16 (March/April), 10-15.
Sherden, W.A., 1988, Gaining the service quality advantage, Williams, F., A.F. Phillips and P. Lum, 1985. Gratifications
Journal of Business Strategy 9 (March-April), 45-48. associated with new communication technologies, in: K.E.
Silpakit, P. and R.P. Fisk, 1985, 'Participating' the service Rosengren et al., eds., Media gratifications research
encounter: A theoretical framework, in: T.M. Bloch et al., (SAGE Publications, Beverly Hills), 241-252.
eds., Services marketing in a changing environment World of Banking, 1986, Vol. 5 (5), Australian bank sets the
(American Marketing Association, Chicago), 117-121. pace for automated branches, 31-32.
Solomon, M.R., C.F. Surprenant, J.A. Czepiel and E.G. Gut- World of Banking, 1990, Vol. 9 (2), The role of technology in
man, 1985, A role theory perspective on dyadic interac- service quality, 12-31.
tions: The service encounter, Journal of Marketing 49 Zeithaml, V.A., 1988, Consumer perceptions of price, quality
(Winter), 99-111. and value: A means-end model and synthesis of evidence,
Surprenant, C.F. and M.R. Solomon, 1987, Predictability and Journal of Marketing 52 (July), 2-22.
personalization in the service encounter, Journal of Mar- Zeithaml, V.A. and M.C. Gilly, 1987, Characteristics affecting
keting 51 (2), 86-96. the acceptance of retailing technologies: A comparison of
Taylor, S., 1994, Waiting for service: The relationship between elderly and nonelderly consumers, Journal of Retailing 63
delays and evaluations of service, Journal of Marketing 58 (1), 49-68.
(April), 56-69.

You might also like