Content uploaded by John Richard Williams
Author content
All content in this area was uploaded by John Richard Williams on Nov 20, 2017
Content may be subject to copyright.
Trust considerations on attitudes towards online purchasing: The moderating effect
of privacy and security concerns
Patrick McCole
a,
⁎, Elaine Ramsey
b
, John Williams
c
a
Queen's University Management School, 25 University Square, Belfast, BT7 1NN, Northern Ireland, United Kingdom
b
Ulster Business School, University of Ulster, United Kingdom
c
School of Business, University of Otago, New Zealand
abstractarticle info
Article history:
Received 1 June 2008
Received in revised form 1 December 2008
Accepted 1 February 2009
Keywords:
Privacy and security
Trust
Online purchasing behavior
The research examines the relationships between three common trust considerations (vendor, Internet and
third parties) and attitudes towards online purchasing. The study incorporates privacy and security concerns
as a moderating variable and finds that these relationships vary depending on the level of concerns a
consumer has when purchasing online. The study suggests that “fears”surrounding the Internet as a place to
do business still hinder the use of it for e-commerce purposes, but that the presence of a reputable agent
might in some manner mitigate this risk. In the context of business to consumer relationships trust in the
vendor is important for the consumer to accept any risk associated with a transaction. Theoretical
implications for online customer behavior theory are also discussed.
© 2009 Elsevier Inc. All rights reserved.
1. Introduction
In business to consumer electronic commerce there is unequivocal
evidence to suggest that trust is important in facilitating electronic
transactions (e.g. Grabner-Krauter and Kaluscha, 2003). Nowadays
the Internet is ubiquitous and is a key facilitator of communication in
business activities and in the daily life of consumers. However, experts
are convinced that the Internet has still some way to go before
attaining its true e-commerce potential (e.g. Gupta and Kim, 2007;
Jahng et al., 2007). Many experts believe the lack of trust between
transacting parties and the system facilitating the exchange is
impeding the speed at which this potential has yet to be realised
(e.g. Dinev et al., 2006).
Four interacting agents are necessary for e-commerce to take
place. They include the technology, the consumer, third parties and
the seller (Kim et al., 2005). The consumer is a key “stakeholder”in
these interactions because it is he who ultimately decides whether a
transaction will take place or not. At the most basic level, a consumer
must decide if he trusts a particular e-vendor and the Internet as a
place to do business (in either a conscious or subconscious manner)
before an online purchase can take place. Another decision that may
be involved is whether third parties can be trusted to “put things
right”should something go wrong.
Although the main effects of these variables have been extensively
examined in previous research (e.g. Grabner-Krauter and Kaluscha,
2003) our study contributes to the e-commerce literature in two
important ways. First we include three e-commerce trust considera-
tions in one model. This is important because our model allows us to
assess the relative contribution of each trust consideration on our key
dependent variable: attitude towards online purchasing. Second we
add to the relatively small body of literature which is beginning to
question the direct impact of trust on selected attitude, intention or
actual behavior in an online context. There is a lack of research
investigating key moderating variables that influence these well
established paths (Fang et al., 2007). Because trust is only necessary
when uncertainty is present in a given exchange (e.g. Mayer et al.,
1995) we propose that the level of trust between the three trust
considerations and attitude to online purchasing is moderated by
innate human perceptions of risk, uncertainty and/or interdepen-
dency which is embodied in privacy and security concerns.
The primary objective of the study is to investigate the moderating
impact of privacy and security concerns on the relationship between
three e-commerce trust considerations (vendor, Internet and third
parties) and attitude towards online buying. Others have also suggested
that this is an area worthy of further research (e.g. Gefen et al., 2003;
Head and Hassanein, 2002; Kimery and McCord, 2002, 2006).
The paper is organised as follows. First a review of the trust
literature is presented. Then, study hypotheses are presented. This is
followed by a discussion of the methods used, data analyses and
results. Thereafter, implications for theory and practice are discussed.
Finally, conclusions and future research avenues are presented.
2. The concept of trust
The countless definitions of trust reflect that it is a multidimen-
sional construct usually associated with such qualities as integrity,
Journal of Business Research 63 (2010) 1018–1024
⁎Corresponding author. Tel.: +353 2890975957.
E-mail address: p.mccole@qub.ac.uk (P. McCole).
0148-2963/$ –see front matter © 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jbusres.2009.02.025
Contents lists available at ScienceDirect
Journal of Business Research
benevolence, empathy, competence, ability and predictability (Gefen
et al., 2003; Lee and Turban, 2001; McKnight et al., 2002; Urban et al.,
1999). Luhmann (1979) even states that trust is the basis of all social
life. In the commercial world, the various trust considerations
fundamentally influence the attitudes and actions taken by buyers
towards sellers (Urban et al., 1999).
Within the many different conceptualisations and definitions of
trust, there is general consensus that “trust exists in an uncertain and
risky environment”(Bhattacharya et al., 1998, p. 461); “the need
[original author's emphasis] for trust only arises in a risky situation”
(Mayer et al., 1995, p. 711); and that “trust is embedded in
uncertainty”(Hardin, 2002, p. 12). The thinking is that due to the
uncertain and complex nature of online transactions, the need for
trust is fundamentally important to mitigate the effects of risk and
uncertainty in online buyer–seller relationships (e.g. Ha and Stoel,
2008; McKnight and Chervany, 2002; Mayer et al., 1995).
3. Trust in what? Trust considerations in internet customer
buying behavior
A consumer must give consideration to whether or not he (1)
trusts the system facilitating the transaction (the Internet); (2) trusts
a particular vendor; and (3) trusts other third parties to safeguard the
exchange before a decision to purchase online is made (Grabner-
Krauter and Kaluscha, 2003; Kim et al., 2005; Urban et al., 1999).
3.1. Trust in the Internet
Ratnasingam et al. (2002, p. 384) state, “Whereas the traditional
notion of trust primarily focuses on trust in a trading partner, trust in
e-business also incorporates the notion of trust in the infrastructure
and the underlying control mechanism (technology trust) which
deals with transaction integrity, authentication, confidentiality, and
non-repudiation.”Lee and Turban (2001, p. 81) state that “…human
trust in an automated or computerised system depends on three
factors: (1) the perceived technical competence of the system, (2) the
perceived performance level of the system, and (3) the human
operator's understanding of the underlying characteristics and
processes governing the system's behavior.”These factors relate to
the perceived ability of the Internet to perform the task it is supposed
to, as well as the speed, reliability and availability of the system. It
might also extend to whether one has a broad knowledge or
appreciation of the World Wide Web itself. A plethora of studies
stress the importance of trust in the Internet medium for e-commerce
to take place (e.g. Eastlick et al., 2006; Einweiller, 2003; Gefen et al.,
2003; Jarvenpaa et al., 1999, 2000; Lee and Turban, 2001; McKnight
et al., 1998, 2002).
3.2. Trust in the vendor
In the context of business to consumer relationships, marketing
literature finds that trust in the transacting vendor is important for the
consumer to accept the risk which is associated with or inherent in a
given transaction. Numerous studies have established that trusting
beliefs strongly influence customers' intention to purchase from online
vendors (e.g. Gefen and Heart, 2006; Jarvenpaa et al., 1999; 2000).
Indeed a lack of trust is a primary reason why consumers do not
purchase (more) from Internet vendors (Gefen and Heart, 2006;
Grabner-Krauter and Kaluscha, 2003; Jarvenpaa et al., 1999; Lee and
Turban, 2001). Decisions regarding the trustworthiness of an e-vendor
may be the result of accumulated transactions in the past (cognitive
trust) or stem from more emotive bases (affective trust) (Lewis and
Weigert,1985). Since trust in a vendor is a necessary precursor to online
buying (Gefen and Straub, 2004; Jarvenpaa et al., 1999; McKnight et al.,
1998, 2002), Internet merchants must display behaviors and cues that
instil consumer trust in their ability, integrity, predictability, and
benevolence when dealing with or serving online shoppers. The
vendor's reputation and brand are instrumental in communicating
such cues to consumers, and are well known determinants of trust (e.g.
Fang et al., 2007, 2008).
3.3. Trust in third parties
Trusted third parties in an Internet buying context include
“institutions and other third party guarantors that actually sell/
provide certificates pledging integrity, ability and intents…[and] in
the e-commerce context this type of trust is more likely to solve
privacy concerns”(Luo, 2002, p. 115). Luo (2002, p. 115–6) believes
certification provided by third parties “can balance the power and
provide the needed trust between the e-vendor and customers”.
Noteberg et al. (1999 cited in Kaplan and Nieschwietz, 2003, p. 98)
found that the presence of any Web seal was more likely to result in an
online purchase than when no seal was present. Gefen et al. (2003)
also found that institution-based structural assurances (e.g. web-
seals) had a positive and significant impact on vendor trust. Other
studies provide different conclusions. For example, Head and
Hassanein (2002, p. 321) “were surprised to find that the general
awareness of seals-of-approval was still relatively low.... [and] from
those that were aware of trust seal programs, less than half…felt that
these recommendations influence their online purchasing behavior.”
Thus the extent to which trust seals are perceived to be effective and/
or the degree to which consumers trust third parties when purchasing
online therefore remains largely unclear.
4. Uncertainty online: privacy and security concerns
Grabner-Krauter and Kaluscha (2003) distinguish between two
types of uncertainty in an online buying context: system-dependent
and transaction-specific uncertainty. They defined system-dependent
uncertainty as that which “...comprises events that are beyond the
direct influence of actors and can be characterized as exogenous.…
[and] relate to potential technological sources of errors and security
gaps…” (p. 785). Transaction-specificuncertainty is defined as “...
endogenous or market uncertainty.... [and] relates to the Internet
merchant and his potential behaviors in the transaction process”
(Grabner-Krauter and Kaluscha, 2003, p. 786). Urban et al. (1999, p. 9)
stated: “If customers do not trust that their personal data will be kept
private and that payment is secured and executed only with
appropriate authorization, they will not use the Internet.”This
statement suggests that online privacy and security concerns are
inextricably linked and when combined, may actually prevent online
purchasing from taking place.
5. Research model
5.1. Theoretical framework and research hypotheses
Similar to other studies in this field (e.g. McKnight and Chervany,
2002; Pavlou, 2003; Vijayasarathy, 2004), we use the Theory of
Reasoned Action (TRA) (Fishbein and Ajzen, 1975) as the guiding
theoretical framework. The TRA implies that beliefs →attitude →
intention →behavior. Our model focuses on the first half of this
framework. In this study we examine the extent to which beliefs
regarding the trustworthiness of a particular vendor, the system itself
(i.e. the Internet) and third parties lead to a favourable attitude
towards buying online in general.
H1. Trust in a vendor has a positive influence on attitude towards
online purchasing.
H2. Trust in the Internet has a positive influence on attitude towards
online purchasing.
1019P. McCole et al. / Journal of Business Research 63 (2010) 1018–1024
H3. Trust in third parties has a positive influence on attitude towards
online purchasing.
The study tests the moderating effects of privacy and security
concerns on each of the paths from the three e-commerce trust
considerations to attitude towards online purchasing. It is believed
that moderating effects exist due to the interdependent nature of
trust and risk (Mayer et al., 1995). Due to the exploratory nature of
the research we do not infer the direction (+/-) of the moderating
variable on each of the paths; simply hypothesize that impacts
exist.
H4. Perceived privacy and security concerns moderates the relation-
ship between (a) trust in a vendor and attitude towards online
purchasing; (b) trust in the Internet and attitude towards online
purchasing; and (c) trust in third parties and attitude towards online
purchasing.
5.2. Control variables
The study includes three control variables in the model. Our first
control variable is experience. Extant theory and research suggest that
variables other than trust exert significant effects on attitude towards
online purchasing. One of these is experience and it has long been
established that those with more exposure to, and longer experience
with, the Internet have a higher probability of purchasing online (e.g.
Blake et al., 2005, p. 1206). Our second control variable is a measure of
perceived channel risk. We include this to cover any residual
uncertainty not already covered by the privacy and security concerns
variable. We anticipate that this relationship will be positive due to
the unique characteristics of the Internet medium (see Clarke, 2001).
Furthermore, because our theory is based on the assumption that the
need for trust only arises in risky situations (Mayer et al., 1995)itis
important that we establish, or at least control for, other external
sources of risk that exist which might influence a consumer's online
buying behavior. Controlling for this will, in turn, make our findings
more robust regarding the effects of different trust considerations on a
consumer's attitude towards online buying at different levels of
privacy and security concerns (see also Jarvenpaa et al. (2000) who
included a similar control variable). Given the changing profile of the
online shopper (e.g. von Abrams, 2008) we include age as our third
and final control variable. We anticipate that age may no longer be an
important or distinguishing factor in forming favourable attitudes
towards online purchasing behavior (see also Sorce et al., 2005). It is
for this reason that we expect the relationship between age and
attitude to be negative and significant.
6. Research methods
6.1. Source of measures for questionnaire survey method
The questionnaire survey method was used to test the research
hypotheses. Measures for Attitude towards Online Purchasing (ATT)
were adapted from Jarvenpaa et al. (1999). Measures for Trust in
Vendor (TV) were adapted from Balabanis et al. (2001),Chakraborty
et al. (2002) and Yoon (2002). Measures for Trust in Third Parties
(TTP) were inspired by Gefen et al. (2003) and Pavlou and Gefen
(2004). Measures for Trust in Internet (TI) were based on the work of
Einweiller (2003). Measures for Experience (EX) were adapted from
Jamal and Naser (2002). Measures for Perceived Channel Risk (PCR)
were adapted from Jarvenpaa et al. (2000). Measures for Privacy and
Security Concerns (PSC) were adapted from Korgaonkar and Wolin
(1999),Chellappa and Pavlou (2002) and Chellappa (2001). Summary
statements of the items used in the questionnaire may be found in
Tables 1 and 2, but a full copy of the instrument is available from the
lead author upon request.
6.2. Measuring privacy and security concerns
Privacy and security concerns have been viewed by some
researchers as “two clearly distinct constructs”(Belanger et al.,
2002:248), whereasothers do not necessarily agree thatthis distinction
is meaningful in studies of Internet buying behavior (e.g. Fang et al.,
2008; Grewal et al., 2004; Zhang and von Dran, 2000). We decided to
treat privacy and security concerns as one construct in the context of
this study for four main reasons. First, Urban et al. (1999) suggested
that privacy and security concerns are inextricably linked. Second,
Zhang and von Dran (2000) labelled privacy and security concerns as a
hygiene factor associated with essential functionality in order for an
online transaction to take place. Third, in relation to Grabner-Krautner
and Kaluscha's (2003)two types of uncertainty, we believe that privacy
and security concerns associated with online purchasing fits into both
uncertainty camps in that privacy and security “mishaps”can result
from both exogenous (e.g. hackers) and endogenous (e.g. misuse of
data in-house) actions. Fourth, Internet shoppers are seldom able to
differentiate between the two. They are often viewed as one and the
same in that to facilitate a secure payment it is assumed that a private
infrastructure must be in place and vice-versa. Privacy and securitymay
be viewed as theoretically separate constructs, but in the context of the
average person's conception of the Internet, they blend into one
because the technical details are beyond the comprehension of most
users. It is for these reasons that we decided not to decouple the
construct for the purposes of the current research.
6.3. The questionnaire pilot process
Although the measurement items were all derived from the
existing literature and thus may be deemed reliable and have
construct validity, the lead author designed a three-phase process to
“perfect”the questionnaire in an attempt to maximise the response
rate (e.g. de Vaus, 1995). First, individual meetings were scheduled
with university colleagues to discuss the questionnaire in terms of
clarity, design and layout. Based on suggestions received a revised
questionnaire was developed and sent to the same individuals for a
Table 1
Descriptive statistics and communalities.
Variable Communality Mean Std
Deviation
Initial Extraction
Trustworthy 0.72 0.76 5.82 1.23
Integrity 0.75 0.81 5.71 1.20
Dependable 0.76 0.83 5.89 1.08
Honest 0.61 0.64 5.90 1.14
Know very little/know a lot 0.71 0.80 4.92 1.36
Inexperienced/experienced 0.72 0.78 4.81 1.49
Uninformed/informed 0.59 0.63 4.84 1.38
Novice/expert 0.63 0.66 4.27 1.42
Good idea 0.65 0.70 5.51 1.17
Appealing 0.69 0.72 5.69 1.23
Like 0.74 0.79 5.57 1.30
Bad idea (r) 0.55 0.56 5.83 1.17
Dependable 0.86 0.89 4.79 1.19
Reliable 0.87 0.94 4.78 1.17
Confidence 0.75 0.77 4.80 1.20
Store information 0.46 0.56 4.56 1.56
Concerned/bothered 0.49 0.57 4.32 1.57
Uncomfortable credit card 0.50 0.55 4.38 1.73
Uncomfortable 0.46 0.48 3.58 1.61
More appealing 0.29 0.32 5.55 1.58
Fiduciary responsibility mechanisms 0.61 0.85 4.34 1.40
Third parties 0.56 0.62 4.32 1.40
Taken advantage 0.28 0.27 2.77 1.45
Risk perception more risk 0.50 0.55 5.20 1.57
Higher potential for loss 0.67 0.81 5.01 1.54
More uncertainty 0.56 0.63 5.16 1.45
No more risk 0.32 0.35 4.90 1.59
1020 P. McCole et al. / Journal of Business Research 63 (2010) 1018–1024
second evaluation. A small number of further suggestions resulted
in a third draft of the questionnaire. Version three was tested on a
small number of university colleagues outside of the lead author's
department. Responses from this phase also enabled the lead
author to check for completeness, response acquiescence as well
as run basic tests on the psychometric properties of the scales
used in the survey. At the end of this stage no further changes were
necessary and the questionnaire was deemed appropriate for
administration.
6.4. The sample
The data used to test the hypotheses were collected from staff
within a large university in the South Island, New Zealand. The
sampling frame consisted of 4500 university personnel selected
from contact addresses on the university Web site. A random
sample of 1500 was generated from this sampling frame (choosing
every third person). Respondents were asked to complete the
questionnaire only if they had prior purchasing experience (bought
at least once online) and that they could answer the questions in
relation to a product or service they had bought for their own
personal use. This overcame the problem of respondents answering
questions relating to any purchases they made online on behalf of
the university.
Of the 1500 questionnaires that were distributed, 99 were
returned due to lack of experience (i.e. had never bought anything
online for personal use) and a further 85 were returned because that
particular individual no longer worked at the university (wrong
address). Further to this, 383 completed (and useable) questionnaires
were returned, representing approximately a 30% overall response
rate. No significant differences in means were found between the
early and late responses at pb0.05 on any of the variables tested
(Armstrong and Overton, 1977).
Summary demographic details for the sample are as follows. The
sample was dominated by above average household income: 21%
earned NZ$40,000 or less; 35% earned between NZ$40,001 and NZ
$70,000; and 45% earned more than NZ$70,001. The sample was also
well educated: 33% held an undergraduate degree and 46% held a
postgraduate degree. Almost 60% of the sample was female. There was a
wide dispersion of age ranges: 19% werein their 20s or younger; 27% in
their 30s; 27% in their 40s; 21% in their 50s; and 6% were 60 years plus.
6.5. Measurement validation
Exploratory factor analysis (using Principal Axis extraction and
Direct Oblimin rotation) was conducted to assess whether items
loaded onto the factors they were expected to. Communalities and
descriptive statistics of the items are shown in Table 1.
All factors with an Eigenvalue greater than 1 were retained (Hair
et al., 1998). In addition, factor loadings of less than 0.35 were
suppressed and items that cross-loaded were deleted. The final factor
solution contained seven interpretable factors accounting for over
66% of the common and unique variance. The rotated factor loading
matrix is shown in Table 2 which also reports the correlation matrix
and reliability coefficients for the constructs in the study. All
constructs have Cronbach's alphas above the 0.70 threshold advo-
cated by Hair et al. (1998, p. 118). All reliability coefficients are
therefore deemed acceptable. On the basis of the above procedures,
we are confident that convergent and discriminant validity has been
established (see also Pavlou, 2003; Yilmaz et al., 2005).
7. Data analysis and results
7.1. Multivariate assumptions
We checked the data to assess whether any assumptions under-
lying multivariate analysis were violated. Specifically, we investi-
gated the normality of the data, constant variance and linearity. No
serious violations were discovered; thus we proceeded with data
analysis.
7.2. Hypothesis testing
Hierarchical moderated regression analysis (HMRA) using SPSS
(v. 15) was used to test the study hypotheses. When testing the
hypotheses relating to the moderating effect of PSC we followed advice
provided by Hartmann and Moers (1999) and Irwin and McClelland
(2001).HMRA“...requires running two regressions, one with the main-
effects-only... and a second with both main effects and the interaction
term…Asignificant interaction effect is confirmed by the statistical
significance of the additional variance explained by the inclusion of the
interaction term (i.e. the significance of the increase in R
2
)”(Hartmann
and Moers, 1999, p. 294). We describe the regression modelling below.
New variables were created to capture the moderating effect of
PSC by multiplying PSC with each of the three trust considerations:
TV, TI and TTP. We checked the Variance Inflation Factor (VIF) and
Tolerance statistics and found that they were all within the acceptable
limits (see Table 3). Regression diagnostic checks indicated that the
residuals were approximately normally distributed and were uncor-
related with the predicted values, however they had non-constant
Table 2
Rotated factor matrix, factor correlation matrix and construct reliabilities.
Factor PSC TV PCR EX TI ATT TTP
Factor loadings
Store information −0.72 −0.04 −0.07 −0.07 −0.03 0.07 −0.08
Concerned/bothered −0.70 0.01 0.03 0.04 −0.01 0.04 −0.02
Uncomfortable
credit card
−0.54 0.00 0.08 0.18 −0.09 0.03 −0.03
Uncomfortable personal −0.43 −0.03 0.05 0.15 −0.15 0.12 −0.03
More appealing −0.41 0.02 0.18 0.05 −0.03 −0.16 −0.06
Integrity −0.02 0.91 −0.02 0.03 −0.01 0.02 −0.01
Dependable 0.03 0.90 0.04 −0.01 0.01 −0.01 0.02
Trustworthy 0.03 0.86 −0.02 0.05 0.00 −0.02 0.02
Honest −0.03 0.79 −0.03 −0.05 0.01 0.01 −0.02
Higher potential loss −0.08 −0.01 0.85 0.02 0.03 0.01 −0.02
More uncertainty −0.05 −0.06 0.75 −0.03 −0.01 −0.02 −0.05
More risk −0.11 −0.02 0.66 0.04 0.07 0.09 −0.03
No more risk (r) 0.12 0.01 0.62 0.01 −0.08 0.00 0.03
Know very little/
know a lot
0.00 −0.03 0.02 −0.90 0.01 −0.01 −0.01
Inexperienced/
experienced
0.13 0.01 0.02 −0.81 0.01 −0.09 −0.03
Uninformed/informed −0.07 0.04 −0.01 −0.80 0.01 0.07 0.11
Novice/expert 0.03 0.01 −0.06 −0.73 0.03 −0.09 −0.04
Reliable 0.01 −0.02 −0.04 0.02 0.99 0.02 −0.04
Dependable 0.00 0.02 0.00 0.05 0.92 −0.05 0.05
Confidence 0.04 0.03 0.03 −0.08 0.80 −0.01 0.04
Appealing −0.05 0.04 0.03 −0.01 −0.03 −0.86 −0.02
Like 0.02 0.02 −0.02 −0.08 −0.03 −0.84 0.01
Good idea −0.04 −0.03 −0.06 −0.01 0.06 −0.78 0.09
Bad idea (r) 0.12 0.01 −0.03 0.02 0.09 −0.66 0.01
Responsibility −0.04 0.03 0.06 0.00 0.08 −0.03 0.91
Third parties −0.05 0.01 −0.01 −0.11 0.02 −0.03 0.75
Taken advantage 0.16 −0.03 −0.09 0.06 −0.04 −0.02 0.41
Cronbach's α0.95 0.92 0.83 0.90 0.80 0.89 0.76
Factor correlations
PSC
TV 0.08
PCR −0.45 −0.14
EX −0.31 −0.14 0.29
TI 0.35 0.26 −0.35 −0.44
ATT −0.24 −0.35 0.11 0.44 −0.46
TTP 0.53 0.16 −0.31 −0.3 0.37 −0.34
1021P. McCole et al. / Journal of Business Research 63 (2010) 1018–1024
variance. Various transformations of both the dependent and
independent variables were explored; however none eliminated the
problem of heteroscedasticity. We also found that omitting unusual
observations (using Cook's Das a measure of “unusual”) did not affect
the parameter estimates or distribution of residuals.
Hence we have to accept this as a limitation of the study in the
following sense: heteroscedasticity inflates the standard errors of
parameter estimates (but does not bias them). We are left with
models that are more likely to be susceptible to Type II errors. For this
reason we elected to use the 10% level when making judgments of
statistical significance. Note however that we still have evidence that
allows us to place some degree of confidence in the un-biasedness of
the parameter estimates.
Two models were estimated. Model 1 assessed the relationship
between each of the three e-commerce trust considerations (TV, TI
and TTP), the control variables (EX, PCR and Age) and the moderating
variable (PSC) on the study's dependent variable (ATT). As expected
TV, TI, TTP, PCR, EX and Age were all significant (with signs as
expected), providing support for H1–H3. PSC was not significant. This
is important for model testing purposes because if there was a direct
relationship between PSC and ATT then claiming that PSC could also
be a moderating variable might be misleading (see Sharma et al.,
1981). Model 2 builds on Model 1 but includes the moderating effects
of PSC on the relationship between each of the trust considerations
(TV, TI and TTP) on ATT. Model 1 has an adjusted R-square value of
42.8%. Model 2 has an adjusted R-square value of 43.5%. A partial
Ftest (see Table 3) shows that Model 2 explains significantly more
variation than Model 1 at the 10% level of significance. Model 2
provides evidence to suggest that PSC moderates the relationship
between (1) TV →ATT at the 5% level of significance and (2) TI →ATT
at the 10% level of significance. The results of the HMRA tests therefore
provide support for H4(a) and H4(b) but not H4(c). It is important to
note that the significance of the main effects (TV, TI and TTP on ATT)
apply when these variables are averaged. Because we know that PSC
moderates some of these relationships (TV →ATT and TI →ATT) and
that a moderating variable changes the regression slope between the
independent variable and dependent variable, it is important to
ascertain how the direct effect changes at different levels of the
moderator. We interpret the moderating impact of PSC as follows: the
relationship between TV and ATT is positive, but increases when
people have higher PSC. Conversely, the relationship between TI and
ATT is positive, but decreases when people have higher PSC.
7.3. Subgroup analysis of non-significant moderating effects
Since the HMRA did not show a significant moderating effect of
PSC on the relationship between TTP →ATT (H4(c)) an alternative
subgroup analysis was performed as previously suggested by Russ and
McNeilly (1995) and more recently by Lee et al. (2005). The total
sample was divided into approximately equal high-medium-low PSC
groups and a one-way ANOVA was performed. The results provide
evidence that although TTP →ATT is positive and significant (see
Table 3) there is a significant difference in the TTP mean scores for
those with high, medium and low privacy and security concerns. Post
hoc Tukey-HSD tests showed that those exhibiting higher PSC
returned significantly lower TTP scores than those with medium or
low concerns (F(2, 380) = 76.2, pb.000). We therefore claim partial
support for H4(c). The next section of the paper discusses the results
of the tests of the hypotheses and provides an explanation for the
main findings of the research.
8. Discussion
8.1. Theoretical implications
The study finds that trust in a vendor, trust in the Internet and trust
in third parties positively influence attitude towards online purchas-
ing. Although all of these relationships have been proven separately in
previous literature (e.g. Grabner-Krauter and Kaluscha, 2003; Jarven-
paa et al., 1999, 2000; McKnight et al., 2002), to the best of our
knowledge, all three have never been tested in the same model before.
The main contribution of our study rests in the moderating impact
that privacy and security concerns (as a manifestation of risk) has on
the three e-commerce trust considerations (vendor, Internet, third
parties) and attitude towards online purchasing. With the exception
of Gefen and Pavlou (2006) and Fang et al. (2007, 2008) relatively few
studies have questioned the direct impact of trust on selected attitude,
intention or behavior in an online context. This is surprising
considering that Irwin and McClelland (2001) suggested that
investigating moderating relationships are important for advancing
marketing theory and Gefen and Pavlou's (2006, p. 12) assertion that
“...it may be an oversimplification to assume that trust always
increases behavioral intentions in a straightforward, linear fashion....”
Why does trust in a vendor become more important when one has
higher privacy and security concerns (H4(a))? One possible explana-
tion is the mechanics of cognitive and affective-based trust. Decisions
regarding the trustworthiness of an e-vendor may be the result of
accumulated transactions in the past (cognitive trust) or stem from
more emotive bases such as our intuition about someone or some-
thing to gauge if they are trustworthy or not (affective trust) (Lewis
and Weigert, 1985). What this implies is that in the context of
business to consumer relationships, trust in the transacting vendor is
vital for the consumer to accept risk associated with, or inherent in, a
given transaction (e.g. privacy and security concern) and that the
vendor's reputation and brand are known determinants of this trust
(e.g. Fang et al., 2007, 2008). It is therefore plausible that trust in a
vendor becomes more important when one has higher privacy and
security concerns because of the cognitive and/or affective bonds that
consumers have with brands. Such bonds provide a signal of the
vendor's integrity, benevolence, empathy, competence, ability and
predictability in online buying situations (Gefen et al., 2003; Lee and
Turban, 2001; McKnight et al., 2002; Urban et al., 1999).
Table 3
HMRA results.
Model Independent
variable
B Std. error Std. βtpVIF
1 Constant 0.43 0.15 2.87 0.00
TV 0.24 0.04 0.25 6.02 0.00 1.11
TI 0.29 0.05 0.30 6.05 0.00 1.59
TTP 0.14 0.05 0.14 2.80 0.01 1.68
PSC 0.00 0.06 0.00 −0.02 0.99 2.07
PCR 0.18 0.05 0.18 3.67 0.00 1.54
EX 0.28 0.05 0.29 5.96 0.00 1.48
Age −0.01 0.00 −0.12 −2.90 0.00 1.14
Adj. R
2
0.43
Fvalue 38.122 0.00
df 7362
2 Constant 0.39 0.15 2.61 0.01
TV 0.22 0.04 0.22 5.28 0.00 1.18
TI 0.34 0.05 0.35 6.57 0.00 1.89
TTP 0.14 0.05 0.14 2.63 0.01 1.75
PSC 0.01 0.06 0.01 0.09 0.93 2.11
PCR 0.19 0.05 0.19 3.88 0.00 1.55
EX 0.28 0.05 0.28 5.93 0.00 1.50
Age −0.01 0.00 −0.12 −2.79 0.01 1.14
PSC×TI −0.11 0.05 −0.11 −2.29 0.02 1.60
PSC×TV 0.09 0.05 0.08 1.84 0.07 1.17
PSC×TTP 0.03 0.04 0.03 0.60 0.55 1.37
Adj. R
2
0.44
Fvalue 27.773 0.00
ΔR
2
.012
Fvalue for ΔR
2
2.512 0.06
df 3359
1022 P. McCole et al. / Journal of Business Research 63 (2010) 1018–1024
The vendor's ability to engender trust in their consumers therefore
is one of the most important elements in e-marketing (Urban et al.,
1999). Enabling this trust has longer term consequences in terms of
e-loyalty and customer repurchasing behavior (Fang et al., 2007,
2008; Johnson and Hult, 2008; Reichheld and Schefter, 2000).
Why does trust in the Internet lessen when one has higher privacy
and security concerns (H4(b))? Internet purchasing activity necessarily
entails primary interactions with the Internet and World Wide Web and
the extent to which consumers trust this computerised medium affects
their overall Internet purchasing behavior(Lee and Turban, 2001,p.80).
Because the Internet may be viewed as a conduit for private data
interchange it makes sense that overall trust in the system lessens with
higher privacy and security concerns, due to external threats from
hackers, the misappropriation of data by companies, as well other
possible threats from other e-agents. Our finding that trust in the
Internetlessens when one has higher privacy and security concernsmay
be explained by consumer decision making processes which involve a
calculative and rational decisionmaking dimension.For example, from a
transaction cost perspective (Williamson, 1985), “costs”may be so high
that conducting business online becomes too dangerous or risky
because there isn't a suitable “governance structure”to protect against
the malevolent actions of others.
Do privacy and security concerns moderate the relationship
between trust in third parties and attitude towards online purchasing
(H4(c))? At best we can only provide partial support for this
hypothesis. We were surprised to find a non-significant moderating
impact in our HMRA model. One would have expected trust in third
parties to increase when privacy and security concerns were higher.
After all, even for a person who has a long trading history and/or can
confidently predict the actions of the other agent(s) involved in the
exchange, he may still rely on institutional bases of trust (e.g. third
parties) to safeguard his investment against possible malevolence
from exogenous (e.g. hackers) and/or endogenous (e.g. misuse of data
in-house) actions.
8.2. Managerial implications
The overarching managerial implication from our study is the need
to build and maintain trust in an online environment. “Fears”
surrounding the Internet as a place to do business still hinder the
use of it for e-commerce purposes but the presence of a reputable
agent might in some manner mitigate this risk. In the context of
business to consumer relationships, trust in the vendor is important
for the consumer to accept any risk associated with a transaction.
Trust in the vendor has previously been confirmed as a necessary
precursor to online buying and repeat buying behaviors (Gefen and
Straub, 2004; Reichheld and Schefter, 2000). In this regard our results
suggest that online vendors should not take trust for granted.
Continued efforts should be made to instil consumer trust in the
online vendor's ability, integrity, predictability, and benevolence as
this may more positively stimulate customer repurchasing behaviors
(Johnson and Hult, 2008; Qureshi et al., 2009).
9. Conclusions
The study investigated the influence of three e-commerce trust
considerations (vendor, Internet, third parties) on attitude towards
online purchasing. The presence of all three e-commerce trust
considerations explained a significantly higher percent of variance
than other models containing only one or two trust considerations
(supporting H1–H3). The main contribution of our study is that these
relationships differ at given levels of a consumer's privacy and
security concerns. On the one hand we found that the relationship
between trust in vendor and attitude towards online purchasing
becomes more important when people have higher privacy and
security concerns (supporting H4(a)). On the other we found that the
relationship between trust in Internet and attitude towards online
purchasing weakens when people have higher privacy and security
concerns (supporting H4(b)). We also found (but only claim partial
support for (H4(c)) that consumer trust in third parties lessens when
their privacy and security concerns rise.
9.1. Study limitations
The study is not without limitations. First, we do not take a cross
cultural approach which somewhat limits the generalizability of our
findings. Second, our results are cross-sectional and therefore only
provide a narrow historical window onto the complex world of online
buying behavior. Third, in order to take part in our study respondents
had to have experience purchasing online. In this regard the findings
in our study might be different for consumers who have never
purchased online before either by wired or wireless means. Fourth,
our study does not explicitly measure or account for an individual's
tolerance/aversion to risk and/or variety seeking behavior/tendencies.
This might have limited our study's potential to uncover unexpected
nuances still to be explored. We urge fellow researchers to take these
issues into account when validating our model in other settings.
9.2. Future research avenues
The research presents some important future research avenues.
Future research should consider undertaking experiments to further
investigate the role of third parties in online buying behavior. Also,
because this work is exploratory in nature, future work should extend
and replicate the study in other settings. Future research should
revalidate the measurement scales developed in this research and test
the relationship/dependencies between the variables in our model. A
lack of qualitative work exists in this field which would yield greater
insights into some of the results presented in this and other studies.
Finally, more research is required to establish important moderators
that affect the direction and/or strength of the relationship between
trust and some other dependent variable such as intended purchase
or actual purchase. For example Gefen and Pavlou (2006) found that
trust on transaction intentions was significant only when respondents
perceived the regulatory effectiveness of online marketplaces to be
“moderate”but not when they perceived it to be either “high”or
“low”. The current study found that the impact of the three e-
commerce trust considerations (vendor, Internet, third parties) on
attitude towards online purchasing differed depending on consumer
privacy and security concerns. The results of both studies suggest that
future research should take into consideration conditions under
which trust positively or negatively influences buying behavior in an
online context.
Acknowledgments
The lead author gratefully acknowledges financial support pro-
vided by the University of Otago, School of Business Research Grant
scheme to conduct this study. He also wishes to thank Eelko Huizingh
for being a ‘critical friend’when called upon for advice during the
early stages of the research project.
References
Armstrong JS, Overton TS. Estimating nonresponse bias in mail surveys. J Mark Res
1977;14(3):396–402.
Balabanis GM, Reynolds N, Simintiras A. Bases of e-store loyalty: perceived switching
barriers and satisfaction. J Bus Res 2001;59:214–24.
Belanger F, Hiller JS, Smith WJ. Trustworthiness in electronic commerce: the role of
privacy, security and site attributes. J Strateg Inf Syst 2002;11:245–70.
Bhattacharya R, Devinney TM, Pillutla MM. A formal model of trust based on outcomes.
Acad Manage Rev 1998;23(3):459–72.
Blake BF, Neuendorf KA, Valdiserri CM. Tailoring new websites to appeal to those most
likely to shop online. Technovation 2005;25:1205–14.
1023P. McCole et al. / Journal of Business Research 63 (2010) 1018–1024
Chakraborty G, Lala V, Warren D. An empirical investigation of antecedents of B2B
websites' effectiveness. J Interac Mark 2002;16(4):51–72.
ChellappaR.K. Consumers'trust in electroniccommerce transactions:the role of perceived
privacy and perceived security. http://www.bus.emory.edu/ram/Papers/sec-priv.pdf,
2001. Accessed 29 May 2008.
Chellappa RK, Pavlou PA. Perceived information security, financial liability and
consumer trust in electronic commerce transactions. Logist Inf Manag 2002;15
(5/6):358–68.
Clarke R. Trust in the context of e-business. http://www.anu.edu.au/people/Roger.
Clarke/EC/Trust.html 2001. Accessed 29 May 2008.
de Vaus DA. Surveys in social research. 4th Ed. St. Leonards, NSW Australia: Allen &
Unwin; 1995.
Dinev T, Bellotto M, Hart P, Russo V, Serra I, Colautti C. Privacy calculus model in
e-commerce —a study of Italy and the United States. Eur J Inf Syst 2006;15:389–402.
Eastlick MA, Lotz SL, Warrington P. Understanding online B-to-C relationships: an
integrated model of privacy concerns,trust, and commitment. J BusinessRes 2006;59:
877–86.
Einweiller S. When reputation engenders trust: an empirical investigation in business-
to-consumer electronic commerce. Electron Markets 2003;13(3):196–209.
Fang Y., Qureshi I., McCole P., Ramsey E. The moderating role of perceived effectiveness
of third party control on trust and online repurchasing intentions. Proceedings,
Americas conference on information systems 2007:1–18.
Fang Y., Qureshi I., McCole P., Ramsey E. Relationship quality and online customer
retention: the contingent role of e-commerce institutional mechanism effective-
ness. Proceedings, Pre-ICIS WeB 2008 7th workshop on e-business: designing e-
business systems: markets, services and networks; Paris, 2008.
Fishbein M, Ajzen I. Belief, attitude, intention, and behavior: an introduction to theory
and research. Reading, MA: Addison-Wesley; 1975.
Gefen D, HeartT. On the need to include national culture as a central issue ine-commerce
trust beliefs. J Glob Inf Manage 2006;14(4):1-30.
Gefen D., Pavlou, P.A. An inverted-U theory of trust: the moderating role of perceived
regulatory effectiveness of online marketplaces. Proceeding, 27th international
conference on information systems 2006:1–19.
Gefen D, Straub DW. Consumer trust in B2C e-Commerce and the importance of social
presence: experiments in e-Products and e-Services. Omega 2004;32:407–24.
Gefen D, Karahanna E, Straub DW. Trust and TAM in online shopping: an integrated
model. MIS Quarterly 2003;27(1):51–90.
Grabner-Krauter S, Kaluscha EA. Empirical research in on-line trust: a review andcritical
assessment. IntJ Human-Comput Stud 2003;58:783–812.
Grewal G, IyerGR, Levy M. Internet retailing: enablers, limiters and marketconsequences.
J Bus Res 2004;57:703–13.
Gupta S, Kim HW. The moderating effect of transaction experience on the decision
calculus in on-line repurchase. Int J Electron Commer 2007;12(1):127–58.
Ha S, Stoel L. Consumer e-shopping acceptance: antecedents in a technology accep-
tance model. J Business Res 2008. doi:10.1016/j.jbusres.2008.06.016.
Hair JF, Anderson RE, Tatham RL, Black WC. Multivariate data analysis. Fifth Edition.
Upper Saddle River, NJ: Prentice Hall; 1998.
Hardin R. Trust and trustworthiness. New York: Russell Sage Foundation; 2002.
Hartmann FGH, Moers F. Testing contingency hypotheses in budgetary research: an
evaluation of the use of moderated regression analysis. Account Organ Soc 1999;24:
291–315.
Head MM, Hassanein K. Trust in e-commerce: evaluating the impact of third-party
seals. Quart J Electronic Commerce 2002;3(3):307–25.
Irwin JR, McClelland GH. Misleading heuristics and moderated multiple regression
models. J Mark Res 2001;38(1):100–9.
Jahng J, Jain H, Ramamurthy K. Effects of interaction richness on consumer attitudes and
behavioral intentions in e-commerce: some experimental results. Eur J Inf Syst
2007;16:254–69.
Jamal A, Naser K. Customer satisfaction and retail banking: an assessment of some of
the key antecedents of customer satisfaction in retail banking. Int J Bank Mark
2002;20(4):146–60.
Jarvenpaa SL, Tractinsky N, Saarinen L, Vitale M. Consumer trust in an Internet store: a
cross-cultural validation. J Comput-Mediat Commun 1999;5(2).
Jarvenpaa SL, Tractinsky N, Vitale M. Consumer trust in an Internet store. Info Technol
Manage 2000;1(1/2):45–71.
Johnson C., Hult, P. Web buyers and their expectations grow up —experienced web
buyers are becoming the new mainstream. Forrest Research, 2008., http://www.
forrester.com/Research/PDF/0,5110,45462,00.pdf. Accessed April 2008.
Kaplan SE, Nieschwietz RJ. A Web assurance services model of trust for B2C e-
commerce. Int J Account Info Syst 2003;4:95-114.
Kim DJ, Song YI, Braynov SB, Rao HR. A multidimensional trust formation model in B-to-
C e-commerce: a conceptual framework and content analyses of academia/
practitioner divide. Decis Support Syst 2005;40:143–65.
Kimery K.M., McCord M. Third party assurances: the road to trust in online retailing.
Proceedings, 35th Hawaii international conference on system sciences 2002. http://
csdl2.computer.org/comp/proceedings/hicss/2002/1435/07/14350175.pdf.
Accessed 29 May 2008.
Kimery KM, McCord M. Signals of trustworthiness in e-commerce: consumer under-
standing of third party assurance seals. J Electron Commer Organ 2006;4(4):52–74.
Korgaonkar PK, Wolin LD. A multivariate analysis of web usage. J Advert Res 1999;39(2):
53–68.
Lee MKO, Turban E. A trust model for consumer internet shopping. Int J Electron
Commer 2001;6(1):75–91.
Lee HY, Lee YK, Kwon D. The intention to use computerized reservation systems: the
moderating effects of organisational support and supplier incentive. J Bus Res
2005;58:1552–61.
Lewis JD, Weigert A. Trust as a social reality. Soc Forces 1985;63(4):967–85.
Luhmann N. Trust and power. London: Wiley; 1979.
Luo G. Trust production and privacy concerns on the internet: a framework based on
relationship marketing and socialexchange theory.Ind Mark Manage 2002;31:111–8.
Mayer RC, Davis JH, Schoorman D. An integrative model of organisational trust. Acad
Manage Rev 1995;20(2):709–34.
McKnight DH, Chervany NL. What trust means in e-commerce customer relationships:
an interdisciplinary conceptual typology. Int J Electron Commer 2002;6(2):35–59.
McKnight DH, Cummings LL, Chervany NL. Initial trust formation in new organizational
relationships. Acad Manage Rev 1998;23(3):473–90.
McKnight DH,Choudhury V, Kacmar C. The impactof initial consumer trust onintentions
to transact with a web site: a trust building model. J Strateg Inf Syst 2002;11(3/4):
297–323.
Pavlou PA. Consumer acceptance of electronic commerce: integrating trust and risk
with the technology acceptance model. Int J Electron Commer 2003;7(3):101–34.
Pavlou P, Gefen D. Building effective online marketplaces with institution-based trust.
Inf Syst Res 2004;15(1):37–59.
Qureshi I, Fang Y, Ramsey E, McCole P, Ibbotson P, Compeau D. Understanding online
customer repurchasing intention and the mediating role of trust —an empirical
investigation in two developed countries. European Journal of Information Systems
2009;18(3):205–22.
Ratnasingam P., Pavlou P.A., Tan Y. The importance of technology trust for B2B electronic
commerce. Proceeding, 15th bled electronic commerce conference 2002. http://
ecom.fov.unimb.si/proceedings.nsf/0/3edd0cb3dfa76aa6c1256e9f0037a3da/
$FILE/ratnasingam.pdf. Accessed 17 November 2008.
Reichheld FF, Schefter P. E-loyalty: your secret weapon on the web. Harvard Bus Rev
2000;78:105–13.
Russ FA, McNeilly KM. Links among satisfaction, commitment and turnover intentions:
the moderating effect of experience, gender and performance. J Bus Res 1995;34:
57–65.
Sharma S, Durand RM, Gur-Arie O. Identification and analysis of moderator variables.
J Mark Res 1981;18(August):291–300.
Sorce P, Perotti V, Widrick S. Attitude and age difference in online buying. Int J Retail
Distrib Manage 2005;33(3):122–32.
Urban, G.L., Sultan, F., Qualls, W. Design and evaluation of a trust based advisor on the
Internet. Working paper MIT. 1999: http://ebusiness.mit.edu/research/papers/
Urban.pdf Accessed 2nd November 2008.
Vijayasarathy LR. Predicting consumer intentions use online shopping: the case for an
augmented technology acceptance model. Inf Manage 2004;41:747–62.
von Abrams K. UK E-commerce Update E-mar keter March 20 08. http://w ww.
emarketer.com/Article.aspx?id=1006650. Accessed 13th November 2008.
Williamson OE. The economic institutions of capitalism. New York: Free Press; 1985.
Yilmaz C, SezenB, Ozdemir O. Joint and interactive effects of trustand (inter) dependence
on relational behaviors in long-term channel dyads. Ind Mark Manage 2005;34:
235–48.
Yoon SJ. The antecedents and consequences of trust in online-purchase decisions.
J Interact Market 2002;16(2):47–63.
Zhang P, von Dran GM. Satisfiers and dissatisfiers: a two way model for website design
and evaluation. J Am Soc Info Sci 2000;51(14):1253–68.
1024 P. McCole et al. / Journal of Business Research 63 (2010) 1018–1024
