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