International Journal of Bank Marketing

Demographic influences on behaviour: An update to the adoption of bank delivery

channels
Ana S. Branca
Article information:
To cite this document:
Ana S. Branca, (2008),"Demographic influences on behaviour", International Journal of Bank Marketing,
Vol. 26 Iss 4 pp. 237 - 259
Permanent link to this document:
http://dx.doi.org/10.1108/02652320810884786
Downloaded on: 11 June 2015, At: 00:29 (PT)
Downloaded by New York University At 00:29 11 June 2015 (PT)

References: this document contains references to 72 other documents.

To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 2360 times since 2008*
Users who downloaded this article also downloaded:
Mark Cleveland, Nicolas Papadopoulos, Michel Laroche, (2011),"Identity, demographics, and consumer
behaviors: International market segmentation across product categories", International Marketing Review,
Vol. 28 Iss 3 pp. 244-266 http://dx.doi.org/10.1108/02651331111132848
Mariëlle E.H. Creusen, (2010),"The importance of product aspects in choice: the influence of
demographic characteristics", Journal of Consumer Marketing, Vol. 27 Iss 1 pp. 26-34 http://
dx.doi.org/10.1108/07363761011012921
Tommi Laukkanen, Vesa Kiviniemi, (2010),"The role of information in mobile banking
resistance", International Journal of Bank Marketing, Vol. 28 Iss 5 pp. 372-388 http://
dx.doi.org/10.1108/02652321011064890

Access to this document was granted through an Emerald subscription provided by emerald-srm:198285 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.

*Related content and download information correct at time of download.

 The current issue and full text archive of this journal is available at
www.emeraldinsight.com/0265-2323.htm

Demographic
Demographic influences on influences on
behaviour behaviour
An update to the adoption of bank delivery
channels 237
Ana S. Branca Received October 2007
Instituto Superior Técnico/CEG-IST, Technical University of Lisbon, Revised January 2008
Accepted March 2008
Lisboa, Portugal

Abstract
Downloaded by New York University At 00:29 11 June 2015 (PT)

Purpose – The purpose of this paper is to examine how demographic characteristics contribute to
consumers’ decision on bank delivery channels’ usage, namely the direct and indirect demographic
influence on channel usage frequency via cognitive and affective mediators.
Design/methodology/approach – The consumer usage frequency pattern concerning the main
bank delivery channels and its determinants are modelled and analysed with a questionnaire sent to
24,000 bank customers. This stage was preceded by a series of in-depth interviews to bank managers
and bank customers.
Findings – Empirical evidence suggests that demographic variables’ influence over consumers’
usage frequency decision has both a direct and indirect component. These influences are identified by
delivery channel.
Research limitations/implications – The main limitation derives from the nature of empirical
results and their generalization to other samples and contexts. Nevertheless, precautions
recommended in the literature to overcome this limitation were followed.
Practical implications – Bank managers will benefit from knowing, by channel, which
demographic characteristics have the desired direct and indirect impact on usage frequency. This
information will improve bank managers’ efforts to encourage customers to favour a specific delivery
channel.
Originality/value – In the literature, only the direct influence of demographics is compared to the
innovation attributes in order to explain the innovation adoption decision. By clarifying the impact of
demographic variables, the paper provides a more robust perception on their role as determinants of
the bank delivery channels’ usage.
Keywords Consumer behaviour, Financial services, Electronic funds transfer systems, Innovation,
Demographics
Paper type Research paper

Introduction
Recent developments in bank delivery channels highlight the importance of
understanding how customers decide to favour a specific delivery channel. This
knowledge is particularly valuable as banks continue to address the financial impact of
new technology-based channels and their success in transferring low value-added International Journal of Bank
transactions to electronic channels. Marketing
Vol. 26 No. 4, 2008
The paper considers usage of the new technology-based delivery channels as an pp. 237-259
innovation in itself, and therefore covers the theories explaining adoption of q Emerald Group Publishing Limited
0265-2323
innovations. The adoption of innovations has been widely researched to understand DOI 10.1108/02652320810884786
 IJBM consumer acceptance and resistance to innovations. When explaining adoption of
26,4 innovations, two usual predictor sets consist of the demographic characteristics and
the perceived attributes of an innovation. Current theories and empirical studies claim
that the perceived attributes of an innovation are better predictors of the extent to
which an innovation is adopted and spread among consumers (Ostlund, 1974; Labay
and Kinnear, 1981; Zeithaml and Gilly, 1987; Rogers, 1995; Rugimbana, 1995; Eastlick,
238 1996). However, such studies depict both demographic and perceived attributes as
direct predictors of adoption. In order to enlighten the adoption process, this paper
explores demographic characteristics as a direct and indirect influence over usage of
technology-based delivery channels. A model is proposed in which the demographic
indirect path of influence occurs via affective and cognitive variables. The affective
and cognitive variables of the model derive from the literature and from in-depth
interviews carried out with bank customers and managers.
The paper begins with a review of innovation adoption theories and proceeds with a
Downloaded by New York University At 00:29 11 June 2015 (PT)

description of the methodology used to investigate the influence of demographic

profiles over consumer usage frequency of the four major bank delivery channels
(ATMs, telephone-based access, internet-based access and the branch network). The
results of this investigation are interpreted and discussed. The paper concludes with
managerial implications.

Adoption of innovations: an overview

Two important concepts concerning the adoption of innovations are the rate of
adoption and the innovativeness behaviour of the target market. Usually, the rate of
adoption determines the financial impact and therefore the success of the innovation,
while innovativeness refers to the degree to which an individual is relatively earlier in
adopting new ideas than other individuals (Rogers, 1995). In the literature, two major
sets of constructs have been hypothesized to contribute to the prediction of
innovativeness and innovation adoption behaviour: perceived innovation attributes
and personal-social characteristics. Several studies on innovations’ adoption focus on
the association between usage of innovations and demographic or psychographic
profiles (Korgaonkar and Smith, 1986; Darian, 1987; Zeithaml and Gilly, 1987; Greco
and Fields, 1991; Rugimbana, 1995; Steenkamp et al., 1999; Lee et al., 2003).
Rogers (1995), when analysing demographic influences on innovativeness,
concluded that early adopters (innovators, early adopters and early majority) are
usually younger, well educated, possessing higher income, low dogmatism and that,
while innovators are more cosmopolitan, the early adopters and the early majority are
more community-bound. Contrastingly, he found that late adopters (including late
majority and laggards) are older, less educated, possessing lower income and higher
dogmatism, local in their social relationships, with limited reliance on outside media
and more reliance on local peers, groups and family. Greco and Fields (1991) in their
analysis of early triers of service innovations, such as an interactive home video
ordering service, found the relevance of income and education (higher levels of income
and education were associated with higher levels of acceptance of home video
ordering), while the impact of age depended on the specific product category being
ordered. Among the demographic variables, Rugimbana (1995) found age (older
consumers were less likely to use ATMs) and occupation (clerical staff were more
 likely to use ATMs than other occupational categories) to be the main predictors of Demographic
consumer ATM usage. Steenkamp et al. (1999) discovered that consumer influences on
innovativeness declined significantly with age. Age was also significant in a study
of the characteristics affecting the acceptance of retailing technologies by Zeithaml and behaviour
Gilly (1987). In their study, it was concluded that the impact of age depended on the
product category: while elderly consumers, compared to non-elderly, were the main
adopters of electronic funds transfer, they were less likely to adopt innovations such as 239
ATMs. Darian (1987) and Korgaonkar and Smith (1986) found that higher education,
higher income, and lower age, while characteristics of earlier adopters, were less
important than in previous years. Lee et al. (2003) found the average computer banking
user to be younger, better educated, and with a higher income compared to the average
ATM user. In a study for US consumers, Kolodinsky et al. (2004) concluded that income
and education positively influenced, and age negatively influenced the adoption of
phone banking and PC banking, while gender was only significant for phone banking.
Downloaded by New York University At 00:29 11 June 2015 (PT)

Despite these studies, there is some scepticism about the role of demographic
variables as determinants of consumers’ preferences and actions: Ostlund (1974) who
studied perceived innovation attributes and personal characteristics as predictors of
innovativeness concluded that perceived innovation attributes were stronger
predictors of new product purchase. This result was also confirmed by Taylor
(1977), Labay and Kinnear (1981), Rugimbana (1995) and Lockett and Littler (1997).
In this paper, the objective is to test the direct and indirect impact of demographic
variables on usage of a bank’s delivery channels. This will clarify the alleged minor
role of demographic variables as predictors of adoption behaviour. The behaviour
studied concerns usage frequency of remote bank delivery channels. The indirect path
is hypothesized to occur via cognitive and affective variables, derived from the
literature and from the in-depth interviews carried out.

Research model
A model is proposed describing consumer behaviour concerning bank service delivery.
According to the model, behaviour is directly explained by affective, cognitive, and
demographic variables. Moreover, demographic variables are proposed to also
indirectly influence behaviour through the intervening affective and cognitive
variables.
Based on the literature and on the in-depth interviews carried out, a list of relevant
affective/cognitive variables and corresponding hypotheses was developed. The
cognitive variables included are perceived service quality, perceived communication
efforts, perceived risk, dependability of technological interfaces and dependability of
human-based interfaces:
.
Service quality is considered to be a cognitive evaluative element that influences
the usage decision for both interpersonal and technology-based encounters
(Gronroos, 1988; Berry and Parasuraman, 1991). In this paper, the scale used to
measure service quality derives from SERVQUAL (Parasuraman et al., 1988) or
more specifically from SERVPERF (Cronin and Taylor, 1992), since it contains
only perception items. But it represents the five dimensions of SERVQUAL,
namely tangibles, reliability, responsiveness, assurance and empathy, adapted to
the banking context and to the focus on technology-based service deliveries.
 IJBM .
Perceived communication efforts were developed to address some of the main
26,4 findings of the in-depth interviews. All customers interviewed mentioned lack of
communication from the banks concerning prices, delivery channels and
satisfaction evaluation. As such, it was decided to include these concerns as an
explaining variable of the channel usage decision. Smith (1989), analysing the
relationships between banks and their small business clients, found that one
240 major element of what was considered to be poor service was insufficient
information. Filotto et al. (1997) found the availability of bank information to be a
“precondition” for having customers aware of the delivery alternatives and to use
them. Other researchers (Mattila et al., 2003; Kuisma et al., 2007) found that the
lack of information caused uncertainty and thus affected Internet banking usage.
.
Perceived risk is particularly relevant in consumer decision making, as it
represents the uncertainty about the potential outcomes of behavior and about
Downloaded by New York University At 00:29 11 June 2015 (PT)

the unpleasantness of those outcomes (Cunningham, 1967; Taylor, 1974; Murray,

1991; Dowling and Staelin, 1994). Perceived risk has been proposed as a predictor
of innovativeness, conceptualised as the degree to which unfavourable outcomes
are perceived as being associated with the innovation (Ostlund, 1974). Generally,
risk aversion has always been seen as an obstacle to innovation adoption
(Darian, 1987). Technology-based channels are innovative and carry some level
of perceived risk (Bhatnagar et al., 2000; Forsythe and Shi, 2003).
.
The dependability of the delivery channels is directly related to the trust and
confidence that consumers have in the channels. For human interaction, it
regards the trust inspired by the employees contacted; and for technology-based
interaction, it concerns the trust and confidence that consumers have that the
delivery interface will perform adequately. The inability of employees to convey
trust to customers will deter customers from using a firm or a delivery channel
characterized by human interaction.
For remote access, the issue is especially appropriate, due to the less familiar context,
compared with human-based interfaces. These variables were considered relevant
following the in-depth interviews carried out, where they were highlighted as
customers’ concerns regarding their relationship with banks:
H1a. A consumer’s age will indirectly influence (through affective and cognitive
factors) that consumer’s usage frequency of a delivery channel.
H1b. A consumer’s gender will indirectly influence (through affective and cognitive
factors) that consumer’s usage frequency of a delivery channel.
H1c. A consumer’s occupation will indirectly influence (through affective and
cognitive factors) that consumer’s usage frequency of a delivery channel.
H1d. A consumer’s education will indirectly influence (through affective and
cognitive factors) that consumer’s usage frequency of a delivery channel.
The affective variables included in the model are enjoyment with technology,
preference for technological interfaces, desire for control, and enjoyment with
participation:
 .
Holbrook and Hirschman (1982) proposed the experiential perspective of Demographic
consumption, in which the consumer is directed by a hedonic pursuit of pleasure, influences on
seeking fun, amusement, fantasy, and enjoyment. More specifically, a number of
researchers have approached the issue of how consumers deal emotionally with
behaviour
technology, namely self-service interfaces. Concerning “enjoyment with
technology”, Langeard et al. (1981) found that some people enjoy playing with
machines and suggested that these people may prefer self-service options. 241
Dabholkar (1996) found a positive relationship between expected enjoyment
from using a technology-based self-service option and expected service quality.
Meuter et al. (2003) suggested that customers have to feel comfortable with
technology in order to use a self-service technology option and that the
technology anxiety construct, focusing on a user’s state of mind about general
technology tools, influenced self-service technologies’ usage. In the information
Downloaded by New York University At 00:29 11 June 2015 (PT)

technology literature, research shows that attitudes have a strong influence on a

person’s adoption of computer technologies (Swanson, 1982, 1988).
.
“Preference for technological interfaces” can be interpreted as the preference for
technological interfaces, as opposed to a preference for human interaction.
Zeithaml and Gilly (1987) found that some people feel strongly that the use of
machines (i.e. technology interfaces) dehumanises service encounters and,
consequently, value their perceived quality of human interaction. Dabholkar
(1996), in a study of influences on the intention to adopt self-service options,
confirmed that the need for interaction with employees was a negative influence
on the intention to use self-service options, mediated by the expected service
quality of the self-service options.
.
Burger and Cooper (1979) introduced the notion of “desire for control” as a stable
personality trait reflecting the extent to which individuals generally are
motivated to control the events in their lives. Motivation for control and the
reaction to the loss of control have been researched in contexts of depression
(Abramson et al., 1978), health (Wallston and Wallston, 1981), academic
performance (Dweck and Licht, 1980) and pain tolerance (Thompson, 1981).
Empirical studies found that consumers choose to participate more extensively
in the delivery of a service because they perceive a stronger sense of control as a
result of the participation (Langeard et al., 1981; Bateson, 1985). Broekhuizen and
Jager (2003) suggested that the relationship between perceived control and
enjoyment is stronger for on-line shoppers compared with off-line shoppers.
.
“Service participation” refers to the degree to which a customer is involved in
producing and delivering the service (Silpatik and Fisk, 1985; Dabholkar, 1990;
Goodwin and Radford, 1993). Bateson’s findings (Bateson, 1985) revealed that
propensity to participate might transcend particular services, and that for certain
individuals “doing it themselves” would be attractive even when the usual
monetary or time-saving incentives were not present. Research results show that
some people find participation intrinsically attractive, while others find more
participative behaviours inherently unattractive (Langeard et al., 1981; Bitner
et al., 1997; Dabholkar, 1996).
 IJBM H2a. A consumer’s age will directly influence that consumer’s usage frequency of a
26,4 delivery channel.
H2b. A consumer’s gender will directly influence that consumer’s usage frequency
of a delivery channel.
H2c. A consumer’s occupation will directly influence that consumer’s usage
242 frequency of a delivery channel.
H2d. A consumer’s education will directly influence that consumer’s usage
frequency of a delivery channel.

Methodology
In order to have a broader and deeper perspective of the research topic, it was decided
to begin with an exploratory method, namely, in-depth interviews. A total of 17
Downloaded by New York University At 00:29 11 June 2015 (PT)

in-depth interviews took place, nine interviews with bank managers and eight
interviews with bank customers. The literature review and the conclusions from the
interviews helped the design of a structured survey questionnaire, by highlighting the
relevant cognitive and affective factors that explain consumers’ usage frequency of
technology-based delivery channels. The questionnaire was distributed to a stratified
random sample of bank customers to collect information on their channel usage
pattern, cognitive and affective variables, and demographic profiles. The sample
comprised four strata of users of the main bank delivery channels: the internet-based
delivery channel, the telephone-based access, the debit-card access (through the ATM
network) and the branch network of the bank. The strata construction followed the
definition of user of a delivery channel as a customer who had used that particular
delivery channel in two specified months of that year (namely, January and February,
considered to be typical months for the banking business).
One major bank agreed to collaborate in the research by distributing the
questionnaire to its customers. In order to have the strata as mutually exclusive groups
of users, it was decided to have a “pyramidal definition” of the strata. Following the
bank rules on access to the delivery channels, the strata were successively extracted
from the global customer base: the first stratum to be identified was users of the
internet-based access, followed by users of the telephone-based channel, debit card
users and branch users. This sequential process avoided duplication of customers
among strata. The branch users received a shorter questionnaire, as questions about
the technology-based channels were not relevant for these customers.
The questionnaire was pre-tested and the final questionnaire was mailed to 6,000
users of each stratum, leading to a total of 24,000 questionnaires sent to users of the
four delivery channels with an overall response rate of 20.2 percent. The questionnaire
comprised closed questions with sections covering usage behaviour patterns (i.e.
delivery channels used and with which frequency), affective and cognitive variables,
and the demographic characteristics of respondents. Usage frequency was defined as
customers’ frequency of use (ranging from “Never”, “Less than once a month”,
“One/two times a month”, “One/two times a week” and “Most days”) of the different
delivery channels. The definition of this dimension of usage follows Ram and Jung
(1991) and Zaichkowsky (1985) who defined usage frequency as how often the product
was used or the different applications for which the product was used. Concerning the
 affective and cognitive variables, respondents were asked about their level of Demographic
agreement with a series of statements (five-point Likert scales). The statements (see influences on
questionnaire items in the following list) were generated following the literature and
the information gathered in the in-depth interviews. Regarding the demographic behaviour
profile, respondents were questioned on gender, age, education and occupation. Gender
was defined as a dummy variable; age was categorised into five classes (ranging from
“18-24 years” to “over 60-years-old”); education and occupation were represented by, 243
respectively, six and seven categorical classes, describing the respondent’s situation.
The statements were:
.
I always prefer to have someone to explain me how to use machines and devices.
.
In XXX, I can choose the delivery channel (branch, ATMs, telephone or internet) I find
more convenient for each contact.
.
It is very frustrating to observe something being done in the wrong way and not being
Downloaded by New York University At 00:29 11 June 2015 (PT)

able to do anything about it.

.
XXX offers me credibility and honesty.
.
Sometimes ATMs are located in dangerous places.
.
If possible, to perform tasks I prefer talking to people rather than use machines.
.
When I try to contact XXX through the ATMs, telephone or internet, the access is always
available and operational.
.
When I decide and control a situation, I feel more confident that it will turn out well.
.
I am used to and known by the employees of my usual branch.
.
When possible, I prefer to do things myself rather than having to wait until someone does
them.
.
I only use the ATMs system during the branches’ opening hours.
.
The Web and ATMs screens of XXX are visually-appealing.
.
Since innovations and technology exist, we better use them.
.
Considering all its delivery channels, XXX has opening hours convenient to all its
customers.
.
My choice between visiting an open branch and using the ATMs system depends on
whether it is raining.
.
The waiting time while the banking transactions are being carried out is insignificant.
.
I enjoy trying different ways of doing the same thing.
.
The XXX employees who I have contact with have the knowledge to answer my
questions.
.
XXX and the employees I contact with are always willing to help.
.
I believe that whatever the banking access I use (branch, ATMs, telephone or Internet), I
have confidentiality guaranteed.
.
XXX gives me suggestions as to how to improve my account management and my usage
of the bank.
.
With the ATMs, telephone or internet, I have always the option of undoing any mistake.
.
Technology eases one’s life.
 IJBM .
I feel distress when going into a bank’s branch full of customers.
26,4 .
At a gas station, I only fill out my car’s tank myself if there is no other alternative.
.
In a XXX branch I receive personalized service.
.
I do not like to use self-service options because I might make some mistake.
.
When someone asks my opinion, it builds my self-confidence.
244 .
I have enough time for leisure activities.
.
XXX delivers the service when promised.
.
In the branch, I always receive the same level of service.
.
XXX performs without errors.
.
When I ask for some service in a XXX branch, the service is provided without
interruptions.
Downloaded by New York University At 00:29 11 June 2015 (PT)

.
There are moments when I feel time-pressured.
.
The XXX employees who I have contact do not ask each other how to carry out specific
operations.
.
Nowadays, technology is everywhere and if I do not adapt, I become outdated.
.
XXX regularly evaluates my satisfaction.
.
I feel safe in the banking operations I have with XXX, whatever the delivery channel used
(branch, ATMs, telephone or internet).
.
XXX employees who I have contact with are always polite and friendly.
.
XXX branches are attractive, modern-looking and functional.
.
Usually I like being able to decide how things will happen.
.
I know that for a cheque book request, XXX charges me differently according to the
delivery channel I use.
.
I am usually pleased to see the things that technology allows us to do.
.
Nowadays, bank channels such as the telephone, the internet and the ATMs are as good
as the branch to do what the customer wants.
.
If I have to wait, I just read or find something to keep me busy.
.
I like having alternative delivery channels to contact XXX.
.
It is humiliating to do something wrong in front of other people.
.
Interacting with technology feels like playing.
.
In my work, I enjoy participating and being involved in the decisions.
.
The weather can influence what I do and how I do it.
.
XXX communicates with me the information that affects me as their customer.
.
XXX offers me personalized service when I use the telephone or the Internet to contact it.
.
XXX gives me prompt service.
.
The complexity I see in the banking channels other than the branch, inhibits me from
using them.
.
When I have contact with XXX through the telephone or the Internet access, I can always
ask for help in case of some difficulty.
 .
I do not like crowded places. Demographic
.
The queues for the ATMs or inside the branches are often large. influences on
.
My choice between going to a branch or to a ATMs machine to withdrawal cash might behaviour
depend on my mood.
.
I trust XXX capabilities to solve problems.
.
When I feel lonely I enjoy talking to someone, even just for a few words. 245
.
Having other people staring at me while I am doing something makes me nervous.
.
XXX is accurate in the information it provides me.
.
If it is raining, I prefer not to leave home.
.
XXX equipment seems up-to-date.
.
When I access XXX and my accounts by a channel other than the branch I always receive
the same level of service.
Downloaded by New York University At 00:29 11 June 2015 (PT)

.
The contact with XXX through the ATMs, telephone or the Internet is easy and with clear
instructions.
.
Due to requests of confidentiality, the bank’s name was replaced by XXX.

Data analysis
The data from the questionnaire were factor analysed to obtain the constructs that
were subsequently related with usage behaviour through ordered logistic regressions.
Factor analysis (with orthogonal Varimax rotation) was used to reduce the items of the
questionnaire to a smaller set of uncorrelated variables representing the cognitive and
affective influences on usage behaviour. These constructs were assessed in terms of
reliability and validity. The most accepted statistic used in marketing to assess
reliability is the Cronbach’s (1951) coefficient alpha, which assesses internal
consistency of a group of items. Cronbach’s alpha for the constructs analysed
ranged from 0.7 to 0.94, with one variable (“desire for control”) holding a lower value of
0.6, but that nonetheless satisfied the threshold (0.5) that Nunnally recommended for
exploratory research (Nunnally, 1967). It was decided to maintain all the variables due
to their theoretical interest, as suggested by Hosmer and Lemeshow (1989). Other
measures of reliability such as test-retest and alternate-form reliability were not
assessed due to the unavailability of the collaborating bank, but also because
test-retest reliability has been criticised as a reliability measure when applied to
perceptual measures (Buchko, 1994). Face validity and content validity were assessed
through the pilot tests and informal discussions of the questionnaire with bank
customers, managers and researchers of the collaborating bank. Discriminant validity
was assessed by seeing if the correlation coefficient between a scale and any other scale
was lower than its Cronbach’s alpha coefficient (Gaski and Nevin, 1985; Andaleeb,
1995). All the constructs presented discriminant validity. Therefore, from the overall
results, it was concluded that all variables were acceptable for face, content and
discriminant validity criteria.
The influence of affective, cognitive and demographic determinants on usage was
analysed with ordered logistic regressions due to the ordered and discrete nature of the
usage variable. The objective was to have the most parsimonious model that best
described the data (Hosmer and Lemeshow, 1989). The model’s goodness-of-fit was
 IJBM assessed with the likelihood ratio test, the Wald test, the deviance, the Akaike
26,4 Information Criterion, and the Bayesian Information Criterion. Wald statistics
sometimes exhibit an unreliable behaviour and are based on a normal distribution,
which often is not followed by categorical constructs (Hauck and Donner, 1977;
Jennings, 1986; Hosmer and Lemeshow, 1989; Agresti, 1990; Agresti, 1996; Menard,
1995). Therefore, as the maximum likelihood procedure is robust to departures from
246 the normality of variables (Jaccard and Wan, 1996), the likelihood ratio test was
preferred to compare and evaluate the models. Care was also taken in relation to
different types of data (numerical) problems (multicollinearity, outliers, influential
observations and complete discrimination) that interfere with model fitting by biasing
the estimates and, consequently, leading to erroneous conclusions.
The model assumes two causal paths feeding into the outcome usage frequency
variable as detailed in Figure 1: the direct impact of the independent demographic
variables and the direct impact of the cognitive and affective mediators. There is also a
Downloaded by New York University At 00:29 11 June 2015 (PT)

path from the demographic independent variables to the mediator variables.

Concerning the structure of the model, tests were carried out to establish whether or
not cognitive and affective variables could be established as mediational variables
between the independent demographic variables and the dependent usage frequency
variable. A variable works as a mediator when it meets the following conditions:
.
variations in levels of the independent variables significantly account for
variations in the presumed mediators;
.
variations in the mediators significantly account for variations in the dependent
variable; and
.
when the described paths are controlled, a previously significant relation
between the independent and the dependent variables is no longer significant,
with the strongest demonstration of mediation occurring when this last path is
zero (Demaris, 1992).

When it is not zero, this indicates the operation of multiple mediating factors. In social
psychology, it is considered that a more realistic goal is to seek mediators that
significantly decrease this last path rather than eliminating altogether the relation
between the independent and the dependent variables (Baron and Kenny, 1986).
To evaluate the structural paths, three regression equations had to be estimated
(Judd and Kenny, 1981; Baron and Kenny, 1986):

Figure 1.
Research model
 (1) the regression of the mediators on the independent variables; Demographic
(2) the regression of the dependent variable on the independent variables; and influences on
(3) the regression of the dependent variable on both the mediators and on the behaviour
independent variables (Table I).

Following the proposed model, these regressions produced the results that are next
presented. First, the regression of the cognitive and affective mediators on the 247
independent demographic variables showed the significant relationships between
demographic and mediating variables by delivery channel.
Second, the regression of the dependent usage variable on the independent
demographic variables, with the corresponding ordered logistic regressions by
delivery channel is detailed in Table II.
Third, Table III illustrates the regression of the dependent usage variable on the
Downloaded by New York University At 00:29 11 June 2015 (PT)

cognitive mediators, the affective mediators and on the independent demographic

variables. By delivery channel, it presents the significant influences on the usage
frequency decision and the corresponding model fit evaluation.

Independent variable and

Delivery channel Dependent variable significance level F-test

Debit-card Pref. tech. interfaces Gender * *, Age * *, Education * *, **

Occupation * *
Depend. tech. interfaces Gender *, Education *
Desire for control Education * * **
Telephone-based Perceived risk Gender *, Education * *
Perceived service quality Age * *, Education * *
Perceived service quality of Age * *, Education * * **
HBSD
Perceived service quality of Education * * **
TBSD
Perceived communication Age * *, Education * *, Occupation * * **
efforts
Depend. of HBSD Education * *
Enjoyment with participation Occupation *
Internet-based Enjoyment with technology Age * * **
Pref. dealing with TBSD Age * *, Education * * **
Desire for control
Perceived service quality Age * *, Education * * **
Perceived service quality of Gender * *, Education * * **
TBSD
Perceived service quality of Age * *, Education * *, Occupation * * **
HBSD
Perceived risk Age *
Branch Perceived service quality Age * *, Education * *, Occupation * **
Tangibles Age * ** Table I.
Perceived communication Education * * ** Regression of the
efforts affective and cognitive
variables on the
Note: * Significant values at 0.05 significance level; * * Significant values at 0.01 significance level independent variables
 IJBM
Variables ÔR SE(ÔR) b 95% CI (ÔR) p-value
26,4
Debit card channel
Gender 1.2 0.18 0.18 (0.89,1.62) 0.24
Age 0.67 0.07 2 0.4 (0.56,0.81) 0.00
Education 1.15 0.06 0.14 (1.05,1.26) 0.00
248 Occupation 0.88 0.03 2 0.13 (0.83,0.93) 0.00
Telephone-based channel
Gender 0.9 0.13 2 0.10 (0.68,1.2) 0.48
Age 1.00 0.07 0.00 (0.87,1.16) 0.99
Education 0.97 0.06 2 0.03 (0.86,1.09) 0.62
Occupation 1.04 0.03 0.04 (0.98,1.10) 0.25
Internet-based channel
Gender 0.70 0.08 2 0.36 (0.57,0.86) 0.00
Age 0.82 0.04 2 0.20 (0.75,0.9) 0.00
Downloaded by New York University At 00:29 11 June 2015 (PT)

Education 1.03 0.05 0.03 (0.93,1.13) 0.64

Occupation 1.02 0.02 0.02 (0.98,1.07) 0.29
Branch channel
Gender 1.07 0.22 0.07 (0.72,1.60) 0.72
Age 1.35 0.19 0.3 (1.03,1.77) 0.03
Education 0.91 0.06 2 0.1 (0.81,1.02) 0.11
Occupation 1.17 0.05 0.16 (1.07,1.28) 0.00
Table II.
Regression of the Note: aThe table presents the estimated odds ratio (ÔR), its standard error (SE(ÔR)), the estimated
dependent variable on the beta coefficient, the 95 percent confidence intervals of the estimated odds ratio and the corresponding
independent variablesa p-value

For significant regressions, mediation implies that the effect of the independent
variables on the dependent variable must be less in the third regression than in the
second regression equation. To have a more complete picture of these relationships, it
was decided to study two mediation models: one according to Baron and Kenny
(1986) and the other according to MacKinnon et al. (2002). Both models are described
in detail.

Mediation model according to Baron and Kenny (1986)

Consistent with Baron and Kenny (1986), the structure of the model can only be tested
for mediations in which the three paths are significant. Moreover, perfect or complete
mediation occurs when, due to control of the mediators, the independent variables have
no effect. Partial mediation occurs if the absolute size of the independent variables’
effect over the outcome variable is reduced, but it is still different from zero when the
mediators are controlled. Then, the test for this mediation involves comparing the beta
coefficient of a demographic variable, as a sole predictor of usage, and the beta
coefficient of the same demographic variable as a predictor of usage, controlling for the
mediator. If the beta coefficient is reduced when the mediator variable is controlled and
the three paths are significant, then it can be said that there is partial mediation. The
variation in the magnitude of the beta coefficient defines the amount of mediation. If
the demographic variable, whose beta coefficient is being compared, is no longer
significant in explaining usage (when the mediator variable is also a predictor) then,
there is full mediation.
 Downloaded by New York University At 00:29 11 June 2015 (PT)

b,c
Delivery channel Variable ÔR/SE(ÔR)/p-value Model fit

Debit-card (731 obs.) Pref. tech. interfaces 1.14/0.11/0.19 LRT(df) 40.14 (6)
Depend. tech. interfaces 1.13/0.09/0.12 p-value 0.00
Desire for control 1.33/0.23/0.09 AIC 2.03
Age 0.77/0.08/0.02 BIC 2 3296.2
Education 1.11/0.06/0.04 Log-likel. 2 732.48
Occupation 0.92/0.03/0.02
Perceived risk 0.97/0.12/0.77
Telephone-based (674 obs.) Perceived service quality 1.07/0.2/0.7 LRT(df) 33.22 (7)
Perceived service quality of HBSD 1.18/0.08 /0.02 p-value 0.00
Perceived service quality of TBSD 1.5/0.2/0.00 AIC 2.36 p-value
Perceived communication efforts 0.82/0.09/0.07 BIC 2 2751.63
Depend. of HBSD 1.17/0.1/0.06 Log-likel. 2 786.58
Enjoyment with participation 1.21/0.11/0.04
Enjoyment with technology 1.17/0.23/0.00
Internet-based (1,467 obs.) Pref. dealing with TBSD 1.14/0.1/0.15 LRT(df) 80.23 (9)
Desire for control 0.94/0.1/0.53 p-value 0.00
Perceived service quality 0.75/0.08/0.01 AIC 2.3
Perceived service quality of TBSD 1.15/0.1/0.1 BIC 2 7253.06
Perceived service quality of HBSD 1.3/0.08/0.00 Log-likel. 2 1677.7
Perceived risk 0.92/0.09/0.39
Gender 0.64/0.07/0.00
Age 0.82/0.04/0.00
Branch (363 obs.) Perceived service quality 2.08/0.49/0.00 LRT(df) 24.34 (4)
Tangibles 0.69/0.11/0.02 p-value 0.00
Perceived communication efforts 0.94/0.14/0.65 AIC 2.16
Occupation 1.13/0.05/0.01 BIC 2 1326.76
Log-likel. 2 385.82
Notes: a The table presents the estimated odds ratio (ÔR), its standard error (SE(ÔR)), the corresponding p-value, and the model fit which includes the
likelihood-ratio test (LRT) with the corresponding degrees of freedom and p-value, the Akaike Information Criterion (AIC), the Bayesian Information
Criterion (BIC) and the log-likelihood of the model. b The likelihood ratio test is performed against the log-likelihood of the intercept-only model; c The BIC
criterion is more adequate for comparing models than just per se. The value of the statistic is reported here for information purposes. It was used for
comparison between models
behaviour

the independent
Demographic

cognitive variables and

both the affective and

demographic variablesa
dependent variable on
Regression of the
influences on

Table III.
249
 IJBM Mediation model according to MacKinnon et al. (2002)
26,4 MacKinnon et al. (2002) propose a less stringent mediation concept that only requires a
significant indirect effect (from the independent variable to the dependent variable via
the mediator or intervening variable). The significance of this indirect effect is assessed
with Goodman’s version of the Sobel test (Goodman, 1960). According to these authors,
mediation does not require a significant influence of the demographic independent
250 variable as a single predictor of usage.
The results for these two perspectives on mediation are summarized in Table IV (by
delivery channel). Concerning the Baron and Kenny’s mediation test, it is reported the
existence or not of mediation: when the mediation is partial, the table shows the
percentage variation in the magnitude of the beta coefficients of the independent
demographic variable and the p-value of the corresponding indirect path. In relation to
the MacKinnon et al.’s mediation definition, the test of the indirect effect was
accomplished by multiplying the beta coefficients of the two regressions involving the
Downloaded by New York University At 00:29 11 June 2015 (PT)

mediator, as referred, and it is described both the significance level ( p-value) associated
with this indirect path and the existence or not of mediation. Other mediations are not
shown due to failure of the significance test in at least one of the required paths, or due
to the fact that the reduction in the beta coefficients was not significant (above the 0.05
threshold).
Regarding the direct influence of the demographic characteristics on usage, it was
additionally tested if the significant demographic variables had to enter the regression
dichotomised, as a set of binary variables, or could be treated as interval variables
without loss of information. Following a suggestion from Long and Freese (2003), a
likelihood ratio test compared the models’ estimates under both options. The results of
this step are not shown due to space restrictions. Nevertheless, the analysis for all
channels revealed that the significant demographic variables could be treated as
interval variables without loss of information.

Discussion
For all channels, some form of mediation (partial or full mediation) through the
relevant affective and cognitive variables (see Table IV). Concerning H1 (i.e. the
indirect influence of the demographic profile on usage frequency), the cognitive and
affective variables were confirmed as significant mediators of age, education, and
occupation for all channels[1]. With the exception of the branch channel, gender’s
influence on usage is mediated by cognitive and affective variables for all channels.
For the telephone-based channel, none of the demographic variables was significant
at explaining usage frequency of the channel. Therefore, as this is a required path to
prove mediation (partial or full), according to Baron and Kenny (1986), the test of
mediation performed followed MacKinnon et al. (2002), which does not require
independent variables to be significant predictors of the dependent variable.
For the internet-based channel, there are two cases in which, although the three
paths are significant, the variation in the beta coefficients is positive (i.e. the beta
coefficient increases) when the mediator is included as a predictor in the regression on
usage. The beta coefficient was expected to decrease (in which case there would be a
partial mediation) or to become insignificant (in which case there would be a full
mediation). The candidate mediations in question are the “enjoyment with technology”
 Downloaded by New York University At 00:29 11 June 2015 (PT)

Delivery channel Mediation testa Independent variable Mediating variable Result Conclusion

Debit-card I Age Pref. tech. interfaces dm b 7.6% ( p ¼ 0.04) Partial mediation

I Education Pref. tech. interfaces d b 19.9% ( p ¼ 0.04) Partial mediation
I Occupation Pref. tech. interfaces d b 7.8% ( p ¼ 0.04) Partial mediation
II Gender Pref. tech. interfaces p ¼ 0.04 Mediation
Telephone-based II Gender Perc. service quality p ¼ 0.03 Mediation
II Age Perc. service quality p ¼ 0.02 Mediation
II Age Perc. serv. qual.HBSD p ¼ 0.01 Mediation
II Education Perc. service quality p ¼ 0.01 Mediation
II Education Perc. serv. qual.HBSD p ¼ 0.02 Mediation
II Education Perc. serv. qual.TBSD p ¼ 0.01 Mediation
II Education Dependability of HBSD p ¼ 0.04 Mediation
II Occupation Perc. serv. qual.HBSD p ¼ 0.046 Mediation
Internet-based I Gender Enjoym. with technology bkb 10.8% ( p ¼ 0.04) Suppression effect
I Gender Perc. serv. qual.HBSD d b 5.8% ( p ¼ 0.04) Partial mediation
I Age Enjoym. with technology db 16.2% ( p ¼ 0.00) Partial mediation
I Age Perc. serv. qual.HBSD bb 14.3% ( p ¼ 0.00) Suppression effect
II Education Pref. tech. interfaces p ¼ 0.031 Mediation
II Education Perc. serv. qual.HBSD p ¼ 0.044 Mediation
II Occupation Perc. serv. qual.HBSD p ¼ 0.006 Mediation
Branch I Age Perc. service quality p ¼ 0.014 Full mediation
I Occupation Perc. service quality d b 16.5% ( p ¼ 0.03) Partial mediation
II Education Perc. service quality p ¼ 0.012 Mediation
Notes: a Refers to the mediation test according to Baron and Kenny (1986), II refers to the mediation test according to MacKinnon et al. (2002)
behaviour

delivery channel
Demographic

Mediation tests by
influences on

251

Table IV.
 IJBM as a mediator of gender’s effect over usage frequency, and the “perceived service
26,4 quality of human-based delivery channel” as a mediator of age’s effect over usage
frequency. According to MacKinnon et al. (2002), a situation in which the magnitude of
the relationship between an independent variable and a dependent variable becomes
larger when a third variable is included indicates suppression. In other words, a
suppression effect would be present when the direct and indirect effects of an
252 independent variable on a dependent variable have opposite signs (i.e. the beta
coefficient of the independent variable, when explaining usage together with the
mediator, is opposite in sign to the product of the beta coefficients for the regressions
involving the mediator). Such cases are known as inconsistent mediation models
(Davis, 1985) and contrast with consistent mediation models in which the direct and
mediated effects have the same sign. It is said that the mediator acts like a suppressor
variable[2]. As such, for the relationships with this suppression effect, it was
considered preferable to calculate the indirect effect and test it for significance (Judd
Downloaded by New York University At 00:29 11 June 2015 (PT)

and Kenny, 1981).

From the results presented on Table IV, one can see that, for example:
.
a fifth of the estimated impact of education on usage frequency of debit card is
accounted for by the intervening variable “preference for technological
interfaces”;
.
the fact of being a male or a female user influences the perception of service
quality which, in turn, influences the frequency with which the telephone-based
access is used: it is not the demographic profile of the user that directly explains
his/her usage of the telephone-based delivery channel, but rather his/her
perception of service quality;
.
16 percent of the estimated impact of age on usage frequency of internet banking
is accounted for by the intervening variable “enjoyment with technology”; and
.
for the branch channel, about 17 percent of the impact of occupation on branch
usage is accounted for by the perception of service quality that the customer has.

Also, for the branch users, findings show that the direct influence of age on usage
totally disappears when the mediator “perceived service quality” is considered (the
indirect effect was confirmed with an associated significance level of 1.4 percent). This
conclusion concerning the impact of age, allows one to understand why, for this
channel, age was not considered a significant predictor of usage. Such a result
illustrates the value of mediation analyses to better capture the influences and the
relationships between variables.
Concerning H2 (i.e. the direct influence of the demographic profile on usage
frequency), Table II shows that usage of the debit card channel decreases with age: an
estimated odds ratio less than 1 indicates that a marginal increase in age will decrease
the odds of more frequent usage of this delivery channel. The results also demonstrate
that the odds of more frequent debit card usage (vs. less frequent) increase with
education, and they decrease with occupation (discrete increase in occupation’s coding
scheme meant going from full-time to part-time employed or from student to retired).
None of the demographic variables were found to be significantly related to usage
frequency of the telephone-based channel. For the internet-based channel, age and
 gender are the significant demographic predictors of usage frequency. Both have Demographic
negative coefficients when regressed over channel usage. The coefficient of 0.82 for age influences on
means that a marginal increase of customer’s age will lower in 18 percent his odds of
using more frequently this channel to contact the bank. This corroborates the literature behaviour
that identifies innovators as younger users. Likewise, in the internet-based channel, the
negative estimated coefficient for gender means that women have lower odds than men
of a more frequent usage vs. a less frequent usage. This result confirms the conclusions 253
of Gefen and Straub (1997) that men are more likely to adopt specific computer
technologies. The branch regression results (see Table II) show that occupation and
age are the only significant demographic influences over usage frequency. In terms of
the demographic direct influence on branch usage frequency, findings illustrate that a
marginal increase in age (i.e. as users get older) increases in 35 percent the odds of a
more frequent usage of the branch. This result is consistent with the situation
experienced by banks that have branches crowded with older customers. Findings
Downloaded by New York University At 00:29 11 June 2015 (PT)

from the branch sample also show the occupation variable with an estimated odds ratio
of 1.17, which means that branch users with higher coding values of occupation (for
example, full time worker compared with part-time worker, or housekeeper compared
with retired) will be 17 percent more likely to visit more often the branch. People who
work less, or who are retired, have more leisure time and so may go more frequently to
the branch. In the in-depth interviews carried out, the need to talk to someone or to feel
occupied was one of the reasons frequently mentioned for having so many older and
retired customers using the branch.

Theoretical contributions
The research objectives focused on customer behaviour, in terms of usage of
technology-based delivery channels. The research literature has repeatedly asked for
contributions in this area (Mick and Fournier, 1998; Steenkamp et al., 1999; Bitner et al.,
2000; Meuter et al., 2000). As hypothesized, the adoption of technology was found to be
indirectly influenced by demographic variables. The demographic profile of a bank
customer affects his adoption behaviour concerning technology-based service
deliveries, but this impact is mediated by cognitive and affective variables. The
discussion on the determinants of innovation adoption is traditionally centred on
innovation attributes (cognitive attributes) and on demographic/personal
characteristics. Between these influences, the literature largely favours innovation
attributes. The results found in this research show that both are significant influences
on adoption/usage, but with different roles: while perceived innovation attributes
represent a direct influence over adoption, demographic factors are a direct and an
indirect influence, with this latter impact mediated by cognitive (perceived channel
attributes) and affective aspects. Therefore, the findings clarify the literature, as:
.
the role of innovation attributes is important, both per se and for the impact of
the demographic characteristics they are mediating; and
.
affective variables explain usage behaviour and also act as mediators of the
demographic profile.

Further, it is now understood that the literature preference for innovation attributes
can be due to a miscomprehension of the phenomenon: it may occur because the
 IJBM perceived innovation attributes comprise a demographic impact, for which they act as
26,4 mediators. The identification of the direct and indirect impact of the demographic
characteristics provides more robust and in-depth information about how the
constructs interrelate.
The role of demographic influences on usage was enlightening, for example,
Stafford (1996) and other researchers concluded that age was a significant
254 discriminator concerning ATM usage. This research has clarified the explaining
power of customer’s age, as its relationship with ATM usage is mediated by cognitive
and affective aspects of the customer. In case of partial mediation, there is a direct and
an indirect path/effect. Therefore, it is important to acknowledge the existence of
mediator variables, as they absorb all or part of the influence of demographic variables.
Traditionally, only the direct effect is considered and studied. Understanding the
complete effect is useful, as demographic profiles represent information that is easily
obtained by companies: customer databases have at least demographic characteristics,
Downloaded by New York University At 00:29 11 June 2015 (PT)

which, as a result, are the most used tool to segment markets. This explains the need to
fully understand their role in affecting consumer behaviour. Conclusions show an
increased importance of demographic profiles as an influence of adoption behaviour
and strongly support the reliance on this type of information to devise marketing
strategies, along with other variables.

Managerial implications
The results from the analysis of the four delivery channels have the potential to offer
new and important insights concerning customers’ usage patterns of the bank’s
delivery channels. The findings clarify the determinants of usage frequency of each
delivery channel, offering valuable information that banks can use to encourage their
customers to favour a specific delivery channel: namely, which demographic
characteristics will have the desired direct and indirect impact on usage frequency and
in which direction.
Bank managers concerned with moving low-value added transactions away from
branches may find some benefit from assuring older users of debit cards of the
dependability of ATMs and other remote channels to provide a satisfying service. By
customer education, bank managers may illustrate the control offered by remote
channels, removing some disbelief that older customers have. An example of valuable
information concerning the telephone-based channel for bank managers and
practitioners consists in the finding that older and/or less educated customers are
more receptive to bank communication efforts. This conclusion could signal these two
groups as having higher priority in the bank’s communication strategy, in order to
optimise their usage of the bank. Care, however, must be taken not to overload the high
value customers with communication. Looking at the demographic characteristics, one
can observe that education is a significant predictor of telephone-based channel users.
In particular, higher educational levels imply higher perceived risk. This result can
help banks to address specifically more educated users of the telephone-based channel
and to communicate them the security procedures and the confidentiality involved in
their interaction with the bank. Likewise, it was found that for Internet users, perceived
risk is a significant predictor of their channel usage frequency. Furthermore, women,
older customers and those highly educated perceive higher risks. This finding can help
 managers to address these customers with more information on security, privacy and Demographic
insurance policies associated with this channel. influences on
behaviour
Limitations and further research
One of the limitations of the study lies in the extent to which the empirical results
obtained can be generalised to other samples and contexts. Although efforts were
undertaken to have a representative sample, by having a large sample size and by
255
choosing the respondents through a stratified random process, the response rate was
low and the analysis concentrated in one bank. Further research could test the model
and replicate its conclusions with other banks or with other service industries.
Another limitation derives from the measures used: other scales from the literature
could have been used to measure the variables in the model. Further research using
different measures and scales could improve the measurement process. Even so, for
Downloaded by New York University At 00:29 11 June 2015 (PT)

some of the constructs included in the model, the literature provides deficient
conceptualisations and measurements, especially for the affective variables. Despite
the limitations, the findings of the study provide a platform/basis for future
investigation and diagnosis: by revealing and highlighting the differences in
customers’ profiles among channels, it has been possible to improve the current
knowledge about consumer behaviour in relation to delivery channel usage.

Notes
1. For the branch channel, the final model only included cognitive variables as significant
predictors of usage frequency: therefore only cognitive variables could be tested as
mediators of the influence of the demographic profile on usage.
2. “A suppressor variable is a variable which increases the predictive validity of another
variable (or set of variables) by its inclusion in a regression equation”, where predictive
validity is assessed by the magnitude of the regression coefficient (Tzelgov and Henik, 1991).

References
Abramson, L., Seligman, M. and Teasdale, J. (1978), “Learned helplessness in humans: critique
and reformulation”, Journal of Abnormal Psychology, Vol. 87 No. 1, pp. 49-74.
Agresti, A. (1990), Categorical Data Analysis, John Wiley & Sons, New York, NY.
Agresti, A. (1996), An Introduction to Categorical Data Analysis, John Wiley & Sons, New York,
NY.
Andaleeb, S. (1995), “Dependence relations and the moderating role of trust: implications for
behavioral intentions in marketing channels”, International Journal of Research in
Marketing, Vol. 12 No. 2, pp. 157-72.
Baron, R. and Kenny, D. (1986), “The moderator-mediator distinction in social psychological
research: conceptual, strategic, and statistical considerations”, Journal of Personality and
Social Psychology, Vol. 51 No. 6, pp. 1173-82.
Bateson, J. (1985), “Self-service consumer: an exploratory study”, Journal of Retailing, Vol. 61
No. 3, pp. 49-76.
Berry, L. and Parasuraman, A. (1991), Marketing Services, Macmillan, New York, NY.
Bhatnagar, A., Misra, S. and Rao, H. (2000), “On risk, convenience, and internet shopping
behavior”, Communications of the ACM, Vol. 43 No. 11, pp. 98-105.
 IJBM Bitner, M., Brown, S. and Meuter, M. (2000), “Technology infusion in service encounters”, Journal
of the Academy of Marketing Science, Vol. 28 No. 1, pp. 138-49.
26,4
Bitner, M., Faranda, W., Hubbert, A. and Zeithaml, V. (1997), “Customer contributions and roles
in service delivery”, International Journal of Service Industry Management, Vol. 8 No. 3,
pp. 193-205.
Broekhuizen, T. and Jager, W. (2003), “A conceptual model of channel choice: measuring online
256 and offline shopping value perceptions”, Working Paper 04F04, University of Groningen,
Groningen.
Buchko, A. (1994), “Conceptualization and measurement of environmental uncertainty:
an assessment of the Miles and Snow perceived environmental uncertainty scale”,
Academy of Management Journal, Vol. 37 No. 2, pp. 410-25.
Burger, J. and Cooper, H. (1979), “The desirability of control”, Motivation and Emotion, Vol. 3,
pp. 381-93.
Cronbach, L. (1951), “Coefficient alpha and the internal structure of tests”, Psychometrika, Vol. 16,
Downloaded by New York University At 00:29 11 June 2015 (PT)

September, pp. 297-334.

Cronin, J. and Taylor, S. (1992), “Measuring service quality: a re-examination and extension”,
Journal of Marketing, Vol. 56 No. 3, pp. 55-68.
Cunningham, S. (1967), “The major dimensions of perceived risk”, in Cox, D. (Ed.), Risk Taking
and Information Handling in Consumer Behavior, Harvard University Press, Boston, MA,
pp. 82-108.
Dabholkar, P. (1990), “How to improve perceived service quality by increasing customer
participation”, in Dunlap, B. (Ed.), Developments in Marketing Science, Vol. 13, Academy of
Marketing Science, pp. 483-7.
Dabholkar, P. (1996), “Consumer evaluations of new technology-based self-service options:
an investigation of alternative models of service quality”, International Journal of Research
in Marketing, Vol. 13 No. 1, pp. 29-51.
Darian, J. (1987), “In-home shopping: are there consumer segments?”, Journal of Retailing, Vol. 63
No. 2, pp. 163-86.
Davis, M. (1985), The Logic of Causal Order, Sage University Paper Series on Quantitative
Applications in the Social Sciences, Sage Publications, Beverly Hills, CA.
Demaris, A. (1992), Logit Modeling, Sage University Paper Series on Quantitative Applications in
the Social Sciences, Sage Publications, Beverly Hills, CA, No. 07-086.
Dowling, G. and Staelin, R. (1994), “A model of perceived risk and intended risk-handling
activity”, Journal of Consumer Research, Vol. 21 No. 1, pp. 119-34.
Dweck, C. and Licht, B. (1980), “Learned helplessness and intellectual achievement”, in Garber, J.
and Seligman, M. (Eds), Human Helplessness: Theory and Applications, Academic Press,
New York, NY, pp. 197-221.
Eastlick, M. (1996), “Consumer intention to adopt interactive teleshopping”, Working Paper
96-113, Marketing Science Institute, Cambridge, MA.
Filotto, U., Tanzi, P. and Saita, F. (1997), “Customer needs and front-office technology adoption”,
International Journal of Bank Marketing, Vol. 15 No. 1, pp. 13-21.
Forsythe, S. and Shi, B. (2003), “Consumer patronage and risk perceptions in internet shopping”,
Journal of Business Research, Vol. 56 No. 11, pp. 867-75.
Gaski, J. and Nevin, J. (1985), “The differential effects of exercised and unexercised power sources
in a marketing channel”, Journal of Marketing Research, Vol. 22 No. 2, pp. 130-42.
 Gefen, D. and Straub, D. (1997), “Gender differences in the perception and use of e-mail: Demographic
an extension to the technology acceptance model”, MIS Quarterly, Vol. 21 No. 4, pp. 389-99.
influences on
Goodman, L. (1960), “On the exact variance of products”, Journal of the American Statistical
Association, Vol. 55, pp. 708-13. behaviour
Goodwin, C. and Radford, R. (1993), “Models of service delivery: an integrative perspective”, in
Brown, S., Bowen, D. and Swartz, T. (Eds), Advances in Services Marketing and
Management, Vol. 2, JAI Press, Greenwich, CT, pp. 231-52. 257
Greco, A. and Fields, D. (1991), “Profiling early triers of services innovations: a look at interactive
home video ordering services”, The Journal of Services Marketing, Vol. 5 No. 3, pp. 19-26.
Gronroos, C. (1988), “Service quality: the six criteria of good perceived quality”, Review of
Business, Vol. 9 No. 3, pp. 10-13.
Hauck, W. and Donner, A. (1977), “Wald’s test as applied to hypotheses in logit analysis”, Journal
of the American Statistical Association, Vol. 72, pp. 851-3.
Downloaded by New York University At 00:29 11 June 2015 (PT)

Holbrook, M. and Hirschman, E. (1982), “The experiential aspects of consumption: consumer

fantasies, feelings, and fun”, Journal of Consumer Research, Vol. 9 No. 2, pp. 132-40.
Hosmer, D. and Lemeshow, S. (1989), Applied Logistic Regression, John Wiley & Sons, New York,
NY.
Jaccard, J. and Wan, C. (1996), Lisrel Approaches to Interaction Effects in Multiple Regressions,
Sage Publications, Thousand Oaks, CA.
Jennings, D. (1986), “Judging inference adequacy in logistic regression”, Journal of the American
Statistical Association, Vol. 81, pp. 471-6.
Judd, C. and Kenny, D. (1981), “Process analysis: estimating mediation in treatment evaluations”,
Evaluation Review, Vol. 5 No. 5, pp. 602-19.
Kolodinsky, J., Hogarth, J. and Hilgert, M. (2004), “The adoption of electronic banking
technologies by US consumers”, International Journal of Bank Marketing, Vol. 22 No. 4,
pp. 238-59.
Korgaonkar, K. and Smith, A. (1986) in Shimp, T., Sharma, S., John, G., Quelch, J., Lindgren, J.,
Dillon, W., Gardner, M. and Dyer, R. (Eds), “Psychographic and demographic correlates of
electronic in-home shopping and banking service”, Proceedings of the 1986 AMA
Educators’ Conference, Series No.52, American Marketing Association, Chicago, IL,
pp. 167-170.
Kuisma, T., Laullanen, T. and Hiltunen, M. (2007), “Mapping the reasons for resistance to
internet banking: a means-end approach”, International Journal of Information
Management, Vol. 27, pp. 75-85.
Labay, D. and Kinnear, T. (1981), “Exploring the consumer decision process in the adoption of
solar energy systems”, Journal of Consumer Research, Vol. 8 No. 3, pp. 271-8.
Langeard, E., Bateson, J., Lovelock, C. and Eiglier, P. (1981), “Services marketing: new insights
from consumers and managers”, Marketing Science Institute Report, No. 81-104.
Lee, E., Lee, J. and Eastwood, D. (2003), “A two-step estimation of consumer adoption of
technology-based service innovations”, The Journal of Consumer Affairs, Vol. 37 No. 2,
pp. 256-82.
Lockett, A. and Littler, D. (1997), “The adoption of direct banking services”, Journal of Marketing
Management, Vol. 13 No. 8, pp. 791-811.
Long, J. and Freese, J. (2003), Regression Models for Categorical Dependent Variables Using Stata,
Stata Press, College Station, TX.
 IJBM MacKinnon, D., Lockwood, C., Hoffman, J., West, S. and Sheets, V. (2002), “A comparison of
methods to test the significance of the mediated effect”, Psychological Methods, Vol. 7 No. 1,
26,4 pp. 83-104.
Mattila, M., Karjaluoto, H. and Pento, T. (2003), “Internet banking adoption among mature
consumers”, Journal of Services Marketing, Vol. 17 No. 5, pp. 514-28.
Menard, S. (1995), Applied Logistic Regression Analysis, Sage University Paper Series on
258 Quantitative Applications in the Social Sciences, No. 07-106, Sage, Thousand Oaks, CA.
Meuter, M., Ostrom, A., Roundtree, R. and Bitner, M. (2000), “Self-service technologies:
understanding customer satisfaction with technology-based service encounters”, Journal
of Marketing, Vol. 64 No. 3, pp. 50-64.
Meuter, M., Ostrom, A., Bitner, M. and Roundtree, R. (2003), “The influence of technology anxiety
on consumer use and experiences with self-service technologies”, Journal of Business
Research, Vol. 56 No. 11, pp. 899-906.
Mick, D. and Fournier, S. (1998), “Paradoxes of technology: consumer cognizance, emotions, and
Downloaded by New York University At 00:29 11 June 2015 (PT)

coping strategies”, Journal of Consumer Research, Vol. 25 No. 2, pp. 123-47.

Murray, K. (1991), “A test of services marketing theory: consumer information acquisition
activities”, Journal of Marketing, Vol. 55 No. 1, pp. 10-25.
Nunnally, J. (1967), Psychometric Theory, McGraw-Hill, New York, NY.
Ostlund, L. (1974), “Perceived innovation attributes as predictors of innovativeness”, Journal of
Consumer Research, Vol. 1 No. 2, pp. 23-9.
Parasuraman, A., Zeithaml, V. and Berry, L. (1988), “SERVQUAL: a multiple-item scale for
measuring consumer perceptions of service quality”, Journal of Retailing, Vol. 64 No. 1,
pp. 12-40.
Ram, S. and Jung, H. (1991), “How product usage influences consumer satisfaction”, Marketing
Letters, Vol. 2 No. 4, pp. 403-11.
Rogers, E. (1995), Diffusion of Innovations, 4th ed., The Free Press, New York, NY.
Rugimbana, R. (1995), “Predicting automated teller machine usage: the relative importance of
perceptual and demographic factors”, International Journal of Bank Marketing, Vol. 13
No. 4, pp. 26-32.
Silpatik, P. and Fisk, R. (1985), “Participatizing the service encounter: a theoretical framework”,
in Block, T., Upah, G. and Zeithaml, V. (Eds), Services Marketing in a Changing
Environment, American Marketing Association, Chicago, IL, pp. 117-21.
Smith, A. (1989), “Service quality: relationships between banks and their small business clients”,
International Journal of Bank Marketing, Vol. 7 No. 5, pp. 28-36.
Stafford, M. (1996), “Demographic discriminators of service quality in the banking industry”,
Journal of Services Marketing, Vol. 10 No. 4, pp. 6-22.
Steenkamp, J.-B., Hofstede, F. and Wedel, M. (1999), “A cross-national investigation into the
individual and national cultural antecedents of consumer innovativeness”, Journal of
Marketing, Vol. 63 No. 2, pp. 55-69.
Swanson, E. (1982), “Measuring user attitudes in MIS research: a review”, OMEGA, Vol. 10 No. 2,
pp. 157-65.
Swanson, E. (1988), Information System Implementation: Bridging the Gap between Design and
Utilisation, Irwin, Homewood, IL.
Taylor, J. (1974), “The role of risk in consumer behavior”, Journal of Marketing, Vol. 38 No. 2,
pp. 54-60.
 Taylor, J. (1977), “A striking characteristic of innovators”, Journal of Marketing Research, Vol. 14 Demographic
No. 1, pp. 104-7.
Thompson, S. (1981), “Will it hurt less if i can control it? A complex answer to a simple question”,
influences on
Psychological Bulletin, Vol. 90 No. 1, pp. 89-101. behaviour
Tzelgov, J. and Henik, A. (1991), “Suppression situations in psychological research: definitions,
implications and applications”, Psychological Bulletin, Vol. 109, pp. 524-36.
Wallston, K. and Wallston, B. (1981), “Health locus of control scales”, in Lefcourt, H. (Ed.), 259
Research with the Locus of Control Construct, Vol. 1, Academic Press, New York, NY,
pp. 189-243.
Zaichkowsky, J. (1985), “Familiarity: product use, involvement or expertise?”, in Hirshman, E.
and Holbrook, M. (Eds), Advances in Consumer Research, Vol. 12, Association of
Consumer Research, Provo, UT, pp. 296-9.
Zeithaml, V. and Gilly, M. (1987), “Characteristics affecting the acceptance of retailing
technologies: a comparison of elderly and non-elderly consumers”, Journal of Retailing,
Downloaded by New York University At 00:29 11 June 2015 (PT)

Vol. 63 No. 1, pp. 49-68.

Corresponding author
Ana S. Branca can be contacted at: anabranca@ist.utl.pt

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com

Or visit our web site for further details: www.emeraldinsight.com/reprints